Qs Weibullalta

Use with Quick Start Repository: Revision 1.
Quick Start Guide
Version 8
Version 8
Weibull++/ALTA 8 Quick Start Guide Part Identification: RPQSWA8 ReliaSoft Corporation Worldwide Headquarters 1450 South Eastside Loop Tucson, Arizona 85710-6703, USA Tel: 1-520-886-0410 Fax: 1-520-886-0399 Sales and Information: 1-888-886-0410 (Toll-free in the U.S. and Canada) ReliaSoft@ReliaSoft.com http://www.ReliaSoft.com 1992-2012 ReliaSoft Corporation, ALL RIGHTS RESERVED. Notice of Rights and Limited Rights to Print this Document If you are a licensed user of the software you are hereby granted the right to print this document in whole or in part, as needed for your exclusive use in conjunction with the use of the software. Except for the limited print rights outlined above, no part of this document may be reproduced or transmitted, in any form or by any means, for any purpose, without the express written permission of ReliaSoft Corporation, Tucson, AZ, USA. Disclaimer Information in this document is subject to change without notice and does not represent a commitment on the part of ReliaSoft Corporation. Companies, names and data used herein are fictitious unless otherwise noted. Use of the software and this document are subject to the terms and conditions set forth in the accompanying license agreement. This software and documentation were developed at private expense; no portion was developed with government funds. Trademarks ReliaSoft, Weibull++, ALTA, Synthesis Platform, Synthesis Elements, BlockSim, Xfmea and RCM++ are trademarks of ReliaSoft Corporation. Other product names and services identified in this document are trademarks of their respective trademark holders, and are used for illustration purposes. Their use in no way conveys endorsement or other affiliation with ReliaSoft Corporation. 10 9 8 7 6
Weibull++/ALTA8 QuickStartGuide
Thank you for your interest in ReliaSoft's Weibull++ and ALTA software tools. This quick start guide has been designed to help you explore many of the software's key features by working through step-by-step instructions for some practical application examples. In order to demonstrate a variety of different applications for the tools available in Weibull++ and ALTA, this guide asks you to imagine that you are a reliability engineer working for a large company with many different product lines, then walks you through the steps you might take to answer a variety of questions from different design groups and customers. Please note that the sample data sets provided are fictional and intended for demonstration purposes. Furthermore, note that although this guide attempts to introduce you to some of the most frequently used tools in the software, Weibull++ and ALTA support many other reliability engineering methods and applications. Within the software, you can choose File > Help to access a wide array of resources that will help you explore other software capabilities. In addition to this introduction, the following chapters are presented in this guide. Note that the examples in chapters 2 through 10 require Weibull++, chapters 11 and 14 can use either ALTA Standard or ALTA PRO and chapters 12 and 13 require ALTA PRO. Chapter 2: Life Data Analysis.............................................................................................................. page 3 Chapter 3: Degradation Analysis ....................................................................................................... page 31 Chapter 4: Design of Reliability Tests ............................................................................................... page 41 Chapter 5: Warranty Data Analysis.................................................................................................... page 57 Chapter 6: Target Reliability Estimation ........................................................................................... page 67 Chapter 7: Stress-Strength Comparison ............................................................................................ page 73 Chapter 8: Competing Failure Modes Analysis ................................................................................ page 81 Chapter 9: Parametric Recurrent Event Data Analysis ..................................................................... page 91 Chapter 10: Event Log Data Analysis ............................................................................................... page 99 Chapter 11: ALTA with One Stress ................................................................................................. page 109
Weibull++/ALTAQuickStartGuide 1
1Weibull++/ALTA8QuickStartGuide
Chapter 12: ALTA with Two Stresses .............................................................................................. page 117 Chapter 13: ALTA with Time-Dependent Stress Profiles ................................................................ page 127 Chapter 14: Accelerated Degradation Analysis ............................................................................... page 135 The quick start repository that is installed with the software (called weibull_alta8_quickstart.rsrp) contains a sample project with completed analyses for all of the examples. To access this file, choose File > Help, click Open Examples Folder, then browse for the file in the Weibull or ALTA sub-folder.
Tip: To preservetheintegrity of theshippedexamplefiles, thesoftwarecreates acopyofthefile eachtimeyou access a repository in the Examples folder. The copy has the same name as the original file and is saved in the default documents folder for your computer (e.g., My Documents\ReliaSoft\Files). Use the copy to work on the exampleprojectsandsaveyourchanges.Anychangesyoumakeinthecopywillnotaffecttheoriginalfile. When applicable, the instructions in this guide will refer to a completed folio in the quick start project using a notation such as (Bulb A - Supplier Data in the sample project). While reading any example in this guide, you have the choice to:

Examine how it was completed in the sample project. Follow the instructions to complete the analyses from scratch. Copy data/analyses from the sample project to help you complete the analyses.
IMPORTANT: Note that it may sometimes be necessary to modify the data in the quick start repository to fit
updated instructions or new examples in the latest printing of this quick start guide. This printing of the guide was designed for use with Revision 1.0 of the Weibull++/ALTA quick start repository. If the file installed on your computer does not display this revision number (visible in the Proprietary Label field in the Project Properties window), the information displayed in the sample projects may not match the instructions printed here. From withinthesoftware,youcanchooseFile>Help>QuickStartGuidetodownloadthelatestprinting.
Weibull++/ALTAQuickStartGuide
LifeDataAnalysis
In life data analysis, the goal is to model and understand the failure rate behavior of a particular item, process or product. The models are built by taking the observed life data, which can be obtained either from the field or from in-house testing. Because time is a common measure of product life, the life data points are often called times-to-failure data. There are two general types of times-to-failure data: complete and censored. In this chapter, you will work with these types of data and use the Weibull++ standard folio to perform life data analysis.
Standardfolios Completeand censoreddata Confidence intervals QCP,plotsand reports
Watchavideoofthisexampleat http://www.ReliaSoft.tv/weibull/startguide/weibull_qsg_2.html.
2.1CompleteData
You receive a request from a team of product engineers who are working on the design of a projector that your company manufactures. You are asked to quantify the life characteristics of the projector bulb in order to understand its reliability. You are given a data set for 10 bulbs that were all tested to failure by the bulbs supplier. The test included shutdown and cooldown periods that were said to simulate normal use conditions. The failure times were 513, 649, 740, 814, 880, 944, 1009, 1078, 1161 and 1282 hours. These failure times are called complete data.
Complete Data: The simplest case of life data is a data set where the failure time of each specimen in the sample is known. This type of data set is referred to as complete data, and is obtainedbyrecordingtheexacttimeswheneachunitfailed.
2LifeDataAnalysis
Objectives

Estimate the average life of the bulb (i.e., MTTF or mean life). Estimate the B10 life of the bulbs (i.e., the time by which 10% of the bulbs will be failed, or the time by which there is a 10% probability that a bulb will fail). Estimate the reliability of the bulbs after 200 hours of operation. Assuming 1,000 bulbs will be fielded, estimate how many would fail after 200 hours. Estimate the warranty time for the bulb if you do not want failures during the warranty period to exceed 2%.
MTTFandMTBF:ThetermsMTBF(meantimebetweenfailures)andMTTF(meantimetofailure) have often been interchangeably used to describe the average time to failure. The truth is, these twometricsarenotthesameandshouldnotbeusedinterchangeably.
When dealing with nonrepairable components (as in the case of life data analysis) the metric sought is the mean time to failure or MTTF. It is only when dealing with repairable components (where the component may fail and be repaired multiple times during its operational life) that you calculatethemeantime betweenfailuresorMTBF.TheonlytimethattheMTTFandMTBFarethe same is when the failure rate is constant (i.e., when the underlying model is an exponential distribution).
Solution
To answer the questions of interest to the design team, you will perform life data analysis on the complete data set provided by the supplier. First, you create a new Weibull++ standard folio by choosing Insert > Folios > Weibull++ Standard Folio.
2.1CompleteData
When prompted to specify the data type, you select Times-to-failure data and clear all the other options. You also use the Units drop-down list to indicate that the time values in the data sheet will be entered in hours, as shown next.
Note:Byclearingalltheotheroptions,youlimitthenumberofcolumnsthatrequiredataentry.Inthisexample,
thereisnoneedtoselectanyoftheotheroptionsbecauseyourdatasetcontainsonlycompletedata.
2LifeDataAnalysis
Once the data folio is created (Bulb A - Supplier Data in the sample project), you enter the data given by the bulb supplier, as shown next.
Note: The new folio starts with single data sheet by default. You can add multiple data sheets to the folio as
needed.ThinkofthedatafolioasanExcelworkbookthatcancontainmultipleworksheets.
Once the data set is entered, the next step is to set up the analysis. The analysis options are available in the Settings area of the control panel. This area provides a quick summary of the analysis settings that will be used to fit a distribution to the data set. The current settings on the Main page of the control panel are as follows:

Rank Regression on X (RRX) Standard Ranking Method (SRM) Median Ranks (MED) Fisher Matrix Confidence Bounds (FM)
2.1CompleteData
These settings are also available on the Analysis page of the control panel, which can be accessed by clicking the Analysis button, as shown next.
Tip: Instead of switching between control panel pages, you can click the blue text in the Settings area on the
Mainpageasaquickmethodforswitchingbetweenavailableanalysisoptions.
The next step is to set the distribution that you want to fit to the data. In order to determine which distribution will work best with your data, you click the Distribution Wizard icon on the Main page of the control panel.
You select to consider all the distributions and then click Analyze to see which of the selected distributions best fits the data set, based on the selected analysis method (in this case, it is for the RRX analysis method). Note:TheDistributionWizardservesonlyasaguide.Youshouldcompareitsresultswithyourownengineering
knowledge about the product being modeled before making the final decision on which distribution to use for yourdataset.
2LifeDataAnalysis
The G-Gamma distribution receives the highest ranking, but you know enough about the properties of that distribution to conclude that it is not the best choice for the type of analysis you are performing. You decide to use the 2P-Weibull distribution, which received the second highest ranking. To do this, you close the Wizard and then select 2P-Weibull from the drop-down list on the control panel. To analyze the data (i.e., fit the selected distribution based on the selected analysis settings), you click the Calculate icon on the control panel.
Once the distribution is fitted, the Analysis Summary area of the control panel displays the parameters of the distribution and other relevant information. In this case, based on the 2P-Weibull distribution setting, the results are as shown next.
This result includes the Rho, which is the correlation coefficient, and the LK Value, which is the value of the log-likelihood function based on the current parameter solution. You can now access additional information, obtain metrics and/or view plots. To create a probability plot, you click the Plot icon on the control panel.
Probability Plots: The Weibull probability plot is drawn on a Weibull probability plotting paper
thatdisplaystimeonthexaxisusingalogarithmicscale,andtheprobabilityoffailureontheyaxis usingadoublelogreciprocalscale.Fromtheplot,youcanobtaintheprobabilityoffailureatagiven timeorviceversa. To obtain the results from the plot, you point to the plot line to highlight it, press SHIFT and then click the plot line. The x and y coordinates for the current position of the mouse pointer will be displayed. You can hold down the SHIFT key while moving the mouse to read the coordinates from any position on the
2.1CompleteData
line. The following example shows the Weibull probability plot with the approximate B10 life of the bulbs (where the x-axis value corresponds to y=10%).
To add a label that displays the coordinates on the plot, you simultaneously press CTRL and ALT, then click a point on the plot, as shown next.
You find that one of the disadvantages of reading results off the plot is that it may be difficult to position the mouse at exact values. To get more precise results, you open the Quick Calculation Pad (QCP) by clicking the icon on the control panel.
2LifeDataAnalysis
The QCP provides quick access to a set of commonly needed results. To estimate the MTTF, you select to calculate the Mean Life and use the Units drop-down list to make sure that the results will be returned in hours. You click Calculate. The average time to failure for the bulbs is about 906 hours, as shown next.
Similarly, you use the QCP to answer the remaining questions:
To estimate the B10 life, you select to calculate the BX% Life and enter 10 for the BX% Life At input. The result is about 570 hours. A B10 of approximately 570 hours means that 10% of the population will be failed by 570 hours of operation. You select Reliability to calculate the reliability at 200 hours. The result is 0.998406 or 99.84%. To determine the number of bulbs out of 1,000 that would fail after 200 hours of operation, you first find the probability of failure at 200 hours by selecting Prob. of Failure and entering 200 for the time. The probability of failure is 0.001594 or 0.16%. You multiply this by the total number of bulbs (1,000). The number of bulbs that would fail is 1.6 or approximately 2. To estimate the warranty time during which no more than 2% of the bulbs will fail, you select to calculate the Reliable Life and enter 0.98 for the Required Reliability input. The result is about 377 hours. (Alternatively, the same result could have been obtained by calculating the B2 life.)
10
2.2CensoredData
2.2CensoredData
After analyzing the suppliers data, you decide that you would feel a lot more comfortable if you could validate the results yourself by performing an in-house test. After putting in the appropriate requests, you are allocated 15 bulbs for testing purposes and have approval to use the lab facilities for a maximum of 40 days. Unfortunately, only five pieces of equipment that are capable of detecting bulb failures are available for you to use. Given this constraint, you decide to test 5 bulbs using the available automated equipment and test the remaining 10 bulbs via manual inspection. You stop by the lab on your way to your office every morning at 8:00 am (except on weekends and holidays) to check the status of the 10 bulbs. When you find that one of the bulbs has failed, the only information you have is that it failed sometime after you last checked on the bulbs (the last inspection) and before now (the current inspection). This is called interval censored data.
Censored Data: Censored data means data with missing information. When a unit has failed
between observations and the exact time to failure is unknown, the time intervals in which the failures occurred are referred to as interval censored data. On the other hand, for nonfailed units that continue to operate after the observation period has ended, the observed operating times of theunitsarereferredtoasrightcensoreddataorsuspensions. Once the test is completed (after 40 days or 960 hours), you have the following data:
For the automated equipment, bulbs 1, 2 and 3 failed at 425, 730 and 870 hours respectively, while bulbs 4 and 5 continued to operate after 960 hours. For the manual inspections, you have the log shown on the following page, where working bulb and x indicates a failed bulb.
indicates
11
2LifeDataAnalysis
Day
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
Hrs
24 48 72 96 120 144 168 192 216 240 264 288 312 336 360 384 408 432 456 480 504 528 552 576 600 624 648 672 696 720 744 768 792 816 840 864 888 912 936 960
Bulb Bulb 6 7

Bulb 8

Bulb 9

Bulb 10

Bulb 11

Bulb 12

Bulb 13

Bulb 14

Bulb 15

Weekend

Weekend

Weekend

Weekend

Weekend

12
2.2CensoredData
Objectives
Use a Failure Rate vs. Time plot to ascertain the failure rate behavior of the bulbs. Your expectation is that the bulbs should have an increasing failure rate. Basically, an increasing failure rate implies an underlying degradation mechanism with time, or in other words, the older the bulb the more likely it is to fail. Estimate the average bulb life. Estimate the B10 life of the bulbs. Estimate the reliability of the bulbs for 200 hours of operation. Assuming 1,000 bulbs will be fielded, estimate how many would fail after 200 hours. Estimate the warranty time for the bulb if you do not want failures during the warranty period to exceed 2%.
Solution
To answer the questions of interest, you will perform life data analysis on the censored data set that you obtained from both parts of the in-house test (automated equipment and manual inspection). First, you create a new Weibull++ standard folio. When prompted to specify the data type, you select Times-to-failure data and the following check boxes:

My data set contains suspensions (right-censored data) My data set contains interval and/or left censored data
13
2LifeDataAnalysis
Once the folio is created (Bulb A - In-House Data in the sample project), you enter the data obtained from both parts of the test in the data sheet. On the control panel, you choose the 2P-Weibull distribution and the MLE parameter estimation method. The resulting data sheet is shown next.
MaximumLikelihoodEstimation(MLE):Asaruleofthumb,analystsoftenusetheMLEanalysis
method when dealing with heavily censored data, as in this example. This is because the MLE method is based on the likelihood function, which considers the timetosuspension data points in the estimate of the parameters whereas in rank regression (RRX and RRY), the solution is based on theplottingpositionsofthetimestofailuredata. Next, you click Calculate to analyze the data set, and then click Plot to create a probability plot.
14
2.3ConfidenceIntervals
On the control panel of the plot sheet, you use the Plot Type drop-down list to choose the Failure Rate vs. Time plot. This displays the failure rate behavior of the bulbs over time. The plot shows an increasing failure rate, as you suspected.
Using the QCP as before, you again answer the rest of the questions:

Mean Life = 894 hours. B10 Life = 593 hours. Reliability at 200 hours = 0.999198 or 99.92%. Probability of failure at 200 hours = 0.000802 or 0.08%. Therefore, when multiplied by 1,000 bulbs, the number of bulbs that would fail is 0.8 or about 1 bulb. Warranty time = 410 hours.
Before you submit the results of the in-house test for the projector bulbs, you want to quantify the error that may be associated with your predictions. After all, the results were based on a sample of only 15 bulbs. If you were to repeat the same experiment with another set of 15 bulbs, you would, in all likelihood, obtain different results. To quantify the error due to sampling, you will use confidence intervals (or
15
2LifeDataAnalysis
confidence bounds), which may be viewed as the range of probable values that you are likely to obtain if you were to keep repeating the same experiment with different samples of 15 bulbs, all belonging to the same homogeneous population.
OneSided or TwoSided Confidence Bounds: Confidence bounds may be onesided or two sided. They are usually around a computed metric, such as the reliability at a given time. Two sidedboundsareusedtoindicatethatthemetricofinterestiscontainedwithintheseboundswith aprobabilityequaltothechosenconfidencevalue.Forexample,90%twosidedconfidencebounds indicatethatthemetricofinterestwillbebetweenthebounds90%ofthetime,andthat5%ofthe time it may be above the upper bound and 5% of the time it may be below the lower bound. Another way to express this is to say that the metric of interest will lie between the bounds with a 90%probability.
Onesidedbounds,on the other hand, are used to indicate that the metricof interestis above the lower bound with a probability equal to the chosen confidence value. For example, a 90% lower onesidedconfidenceboundindicatesthatthemetricofinterestwillbehigherthan thelowerbound witha90%probability.
90%
5% X 5% Y 10%
90%
90%
10% X
Upper 1-sided confidence bounds
X
Lower 1-sided confidence bounds
2-sided confidence bounds
Anestimatedmetricthatisobtainedwithouttheuseofconfidenceboundsissaidtobeatthe50% confidencelevel,ortheerrorisequalonbothsides. You use the QCP to calculate the MTTF of the bulb. To obtain the desired confidence bounds, you choose Two-Sided from the Bounds drop-down list and enter 0.90 for the Confidence Level input. The
16
calculated results are shown next.
Therefore, with a 90% probability, the true value of the MTTF is estimated to be between 790 and 1,010 hours. While unlikely, there is still a 5% probability that the MTTF is less than 790 hours and a 5% probability that it is greater than 1,010 hours.
17
2LifeDataAnalysis
To visualize the confidence bounds, you create a Weibull probability plot. To show the bounds, you rightclick the plot sheet and select Confidence Bounds. You then select to display the Two-sided bounds. The following plot shows the 90% two-sided confidence bounds on the Weibull probability plot.
2.4ComparingDataSets
After analyzing the data from the bulb supplier and the in-house test, a reasonable question would be, how do they compare? Specifically, you want to understand how the data and analysis from the in-house test compare to the data and analysis provided by the bulb supplier. One way to do this is to use an overlay plot that displays both data sets and analyses on the same plot. You create the overlay plot by choosing Insert > Reports and Plots > Overlay Plot.
When prompted to select which data sets to plot, you select the folio containing the data set from the supplier and the folio containing the data set from the in-house test. (Bulb A - Supplier Data and Bulb A - In-House Data in the sample project.)
18
2.4ComparingDataSets
The following overlay plot (Bulb A - Supplier vs. In-House in the sample project) shows the two data sets plotted on the same Weibull probability plot (and both with 90% two-sided confidence bounds):
The overlay plot shows that the probability lines of the two data sets have similar slopes. Furthermore, there is a significant overlap in the confidence bounds of both sets, which suggests that both data sets exhibit the same reliability characteristics. Another way to determine whether there is a statistically significant difference between the two data sets at a given confidence level (say 90% confidence level for this example), is to switch the plot type to a Contour Plot. When prompted to specify the contour lines, you select the 2nd Level, 90% check box. Tip:Toaddcontourlines(confidencelevels)totheplotorchangetheconfidencelevelassociatedwithacontour line,clicktheContoursSetuplinkonthecontrolpaneloftheplotsheet.
19
2LifeDataAnalysis
You add labels to the plot by holding down the CTRL key and then clicking in the plot. You then edit the text by double-clicking the labels. To position the text on the plot, you click the label and then drag it by its handle (i.e., the small box on the upper left corner of the label), as shown next.
The annotated plot is displayed next. The contour plot shows that the two data sets overlap at the 90% confidence level; therefore, they do not show a statistically significant difference at that confidence level.
Tip:FormoreextensiveannotationsyoucanusetheRSDrawutility,whichisametafilegraphicseditorthat allows you to insert text, draw objects, paste another picture into the plot and rearrange the objects in the plot.YoucanopentheutilitybyclickingtheRSDrawicononthecontrolpanel.
20
2.5Reports
2.5Reports
For presentation to the rest of the team, you will prepare some reports that summarize the results of your analyses. Each of the built-in reporting tools in Weibull++ analysis workbooks and Word report templates offers different advantages. You will use both tools for this situation.
Objectives
Use an analysis workbook to automatically calculate all of the results that you obtained individually from the QCP. Use a Word report template to create a document that presents some key results from both analyses (MTTF and B10 life) along with a plot of the failure rate vs. time. Create another sheet in the analysis workbook that can be used to evaluate the expected number of failures for different time intervals, and determine which time interval will have the greatest number of failures.
2.5.1AnalysisWorkbook
To create an analysis workbook that contains all of the individual QCP results automatically calculated in the same worksheet, you choose Insert > Reports and Plots > Analysis Workbook.
When prompted to specify a default data source, you choose the folio containing the bulb suppliers data (Bulb A - Supplier Data in the sample project) and click OK to create a blank workbook.
21
2LifeDataAnalysis
You build a worksheet like the one shown next (Analysis Workbook - Bulb A in the sample project) by using the Function Wizard to build functions that automatically insert calculated results based on the specified data source.
For example, to build the function that is highlighted in the picture, you open the Function Wizard by clicking the icon on the control panel.
22
2.5Reports
You select the AMEAN function from the function list, as shown next.
The AMEAN (i.e., MTTF) function doesn't have any required inputs, and the confidence level input is optional (as indicated by the brackets that enclose the label) and not applicable in this example. So you simply specify the data source by selecting the Use Default check box. From the Data Source Index drop-down list, you select the number 1 (i.e., the first default data source, which you selected when you created the workbook), then you click Insert. The MTTF for the specified data set will be automatically computed and displayed in the workbook. You continue to use the Function Wizard to build the rest of the report:
To calculate the B10 life, you select the TIMEATPF function and enter 0.1 for the Unreliability input. To calculate the reliability at 200 hours, you select the RELIABILITY function and enter 200 for the Age input. To calculate the expected number of failures after 200 hours of operation, you first calculate the probability of failure and then multiply that number by the total number of bulbs (1,000 bulbs in this example). To do this, you enter the following formula directly into the appropriate cell in the analysis workbook: =(1 - B7)* 1000, where B7 is the cell reference of the reliability at 200 hours,
23
2LifeDataAnalysis
as shown in the picture on page 22.
To calculate the B2 life (which is equivalent to the warranty time during which no more than 2% of the bulbs will fail), you select the TIMEATPF function and enter 0.02 for the Unreliability input.
Once you have completed the worksheet, you save it as a template that can be used in the future for any other analysis. To do this, you choose Workbook > Worksheet > Save Workbook as Template.
The template file (*.wrt) is now stored in the location you specified and will be accessible from within the Analysis Workbook Wizard each time you create a new workbook. For example, to create a workbook that shows the results from the in-house test data, you choose Insert > Reports and Plots > Analysis Workbook to open the Wizard again. This time, you select the in-house data folio (Bulb A - In-House Data in the sample project) as the default data source. You then select the Based on Existing Template check box and click the User tab. You click the Open button and browse for the *.wrt file you just saved. (Note that if the template had been opened recently on this computer, the path and filename would be displayed in the list and you could simply double-click the name to open it again.)
When you click OK, the new workbook contains all of the analysis results automatically calculated for the in-house data set.
24
2.5Reports
2.5.2WordReportTemplate
To create a document that presents key results and plots for both data sets, you choose Insert > Reports and Plots > Word Report Template.
When prompted to specify a default data source, you choose the folio containing the bulb suppliers data (Bulb A - Supplier Data in the sample project) and click OK to create a blank workbook. When the template opens, you click the Associate Data Sources icon on the control panel so you can add the in-house data as the second default data source (each report can be associated with up to four data sources at a time).
The Associate Data Sources window is shown next, with the in-house data set already added as a second data source.
You build a report template like the one shown next (Word Report - Bulb A in the sample project) by using the Function Wizard and Plot Wizard to insert functions into the desired locations. You also use
25
2LifeDataAnalysis
some of the more flexible formatting options that this tool provides for enhancing the presentation quality of the report.
For example, to insert the name of the standard folio/data sheet that the results were obtained from (i.e., the [DATASOURCENAME(1)] function in the sample report), you open the Function Wizard by clicking the icon on the control panel.
26
2.5Reports
You select the Data Source Name function from the function list. You also select the number 1 for the data source index. You then click Insert to add the function to the report template. You continue to build the report using a similar process for the other functions. To insert the Failure Rate vs. Time plot, you open the Plot Wizard by clicking the icon on the control panel.
You select the Failure Rate vs. Time plot from the plot list. You also select the number 1 for the data source index. You then click Insert to add a placeholder for the plot in the report template. To resize a plot, you select its placeholder in the report template, and then click and drag the resize handles (i.e., the small circles that appear within the corners of the selected placeholder). When you have finished building the report template, you generate the report by clicking the Create Report icon on the control panel.
This creates a Microsoft Word document with the data inserted in the report, as shown next.
27
2LifeDataAnalysis
2.5.3BuildingTablesandUsingCustomFunctions
To evaluate the expected number of failures for different time intervals, you create another worksheet in the analysis workbook (Analysis Workbook - Bulb A in the sample project), by choosing Workbook > Worksheet > Insert Sheet.
You double-click the name of the new worksheet (located at the bottom of the window) and rename Sheet1 to Expanded Workbook. Next, you build a worksheet like the one shown next by building a table and creating custom functions.
The results in the table show that for a sample size of 1,000 bulbs, the interval from 900 to 950 hours will have the greatest number of failures, about 76.
28
2.5Reports
The following picture shows the formulas in the worksheet (by selecting the View Formulas check box on the control panel).
For example, in the Time column, instead of manually entering the times at 50 hour intervals, you enter a start time in cell A7 and then enter a formula directly in cell A8 that automatically adds 50 hours to the previous time (e.g., =A7 + 50 where A7 is the cell location of the start time). You then copy this formula by clicking inside cell A8 and then dragging the fill handle (the small black square on the lower right of the cell) down the column of cells. In the Reliability column, you use the function wizard to insert a function that calculates the reliability at an interval (e.g., "=RELIABILITY(Default1,A7)" where A7 is the cell location of the time that the reliability will be calculated for). You copy this function down the column of cells, as well. In the Cumulative Number of Failures column, you build a custom function to calculate the cumulative number of failures by the end of each interval for a given sample size. You use the function wizard to insert the ROUND function to round the result to 2 decimal places. You then modify the input for the
29
2LifeDataAnalysis
function directly in the cell (e.g., "=ROUND((1-B7)*$B$4,2)" where "1-B7" gives you the probability of failure for the value calculated in B7, and the absolute reference to cell $B$4 supplies the sample size). Tip: An absolute reference points to a cell at an exact location. When the cell containing the function is copied, thereferencedoesnotchange.Absolutereferencesaredesignatedbyplacingadollarsign($)infrontoftherow and/orcolumnthatistobeabsolute. In the Number of Failures per Interval column, you enter a formula directly in the cell to calculate the number of failures by interval. (e.g., "=C8-C7" where C8 is the cumulative number of failures after 150 hours and C7 is the cumulative number of failures that had already occurred by 100 hours).
30
DegradationAnalysis
Degradation analysis is useful for tests performed on highly reliable products that (a) cannot feasibly be tested to complete failure under normal operating conditions, (b) are associated with a measurable performance characteristic (e.g., the wear in brake pads) and (c) are not destroyed upon inspection. This analysis consists of two steps. First, the failure times of the units on test are extrapolated using measurements of their degradation over time. (A unit is considered failed when its degradation reaches a specified critical level.) Second, once these failure times are obtained, life data analysis is used to estimate the reliability of the product.
Degradation folios Analyzing extrapolated failuretimes QCPandplots
In this chapter, you will use a degradation analysis folio to analyze the measurements of an LED lamps decreasing light output. Watchavideoofthisexampleat http://www.ReliaSoft.tv/weibull/startguide/weibull_qsg_3.html.
3.1SimpleDegradationAnalysisUsingLuminosityMeasurements
After completing the life data analysis for the projectors incandescent bulb, you learn that the R&D group is working on perfecting an LED lamp to replace the bulb. The product engineers would like to obtain reliability information for the new lamp. However, while it is relatively easy to test incandescent bulbs to failure (given their low life in hours), testing LEDs to failure is not feasible given their long life characteristics. Based on the failure mode under consideration for the LEDs, you know that the failure is gradual and preceded by a decrease in light output of the lamp over time. As a result, the R&D group concludes that in lieu of a long failure test, it is preferable to perform a degradation analysis where they will monitor the decrease of light output over time and use this information to extrapolate failure times.
31
3DegradationAnalysis
According to the formal test plan, the light intensity of 10 lamps will be recorded at 200, 600, 800 and 1,000 hours of operation. This data set will be used to predict the lamps life. Furthermore, the engineers agree that the lamp should be considered failed when its output reaches 50% of its design output of 1,000 lumens. The following measurements were obtained in the test: LuminosityMeasurements(inLumens)of10LEDLamps Inspection Time(Hr)
400 600 800 1,000
Lamp 01
960 890 850 810
Lamp 02
950 900 860 820
Lamp 03
990 890 830 780
Lamp 04
840 730 670 615
Lamp 05
880 760 710 670
Lamp 06
800 770 730 700
Lamp 07
930 890 840 820
Lamp 08
870 810 760 730
Lamp 09
870 800 730 690
Lamp 10
910 850 800 770
DegradationAnalysisAssumptions:Degradationanalysiscanbeusedforanalyzingthisdataset because, according to the underlying physics, the luminosity will continue to decrease with time. The main assumption in this kind of analysis is that the degradation mechanism and observed measurementswillcontinuethesamebehaviorthroughoutoperation.Inotherwords,thisanalysis assumesthatthemeasurementswillmonotonicallydecrease(orincrease)withtime. Objectives

Use a Degradation vs. Time plot to see how the luminosity of the lamps degrades over time. Estimate the following metrics:

The average lamp life (i.e., MTTF or mean life) The B10 life (i.e., the time by which 10% of the lamps will be failed) The reliability for 200 hours of operation The number of lamps (out of 1,000) that can be expected to fail after 200 hours The warranty time on the lamp if you do not want failures during the warranty period to exceed 2%
32
Solution
You start by choosing Insert > Folios > Weibull++ Degradation to create a new degradation analysis folio. Then, you enter degradation measurements into the folios data sheet and enter 500 for the lamps Critical Degradation (i.e., the luminosity level at which a lamp is considered failed). First, you need to specify how failure times will be extrapolated from the luminosity measurements. The failure mode under consideration is metal fatigue, and you know the underlying degradation model for this failure mode is the power model, so you select Power from the drop-down list in the Degradation Model area. Next, you need to specify how the extrapolated failure times will be analyzed. Based on your engineering knowledge, you select 2P-Weibull from the Life Data Model area. In the Settings area of the control panel, you click the blue text to switch between the available analysis options. (Note that these settings are also displayed on the Analysis page.) You select the following:

Next, you click the Calculate icon to analyze the data.
33
The folio (Degradation - LED Lamp in the sample project) appears as shown next.
Before you continue, you decide to quickly review the results of the softwares calculations. To see how the failure times were extrapolated, you click the Degradation Results (...) button. You look at the
34
Degradation Fit Results tab (shown next) to see the parameters of the degradation model that was fitted to each lamp.
35
Then you look at the Extrapolated Failure/Suspension Times tab to see the failure times that were extrapolated, as shown next.
36
To see the results of the life data analysis that was performed on the extrapolated failure times, you return to the control panel and click the Life Data Results (...) button.
Next, to view a plot that shows how the luminosity for each unit on test degraded over time, you click the Plot icon on the control panel and select the Degradation vs. Time (Linear) plot. This shows you the degradation lines for each LED lamp. To improve the appearance of the plot, you clear the check boxes in the Scaling area and enter the following ranges:

Y-axis: 450 to 1,000 lumens X-axis: 300 to 3,000 hours
37
The resulting plot is shown next. The horizontal line towards the bottom of the plot marks the luminosity level at which the lamp is considered failed.
Next, to solve for the requested reliability metrics, you open the Quick Calculation Pad by clicking the QCP icon on the control panel.
38
You select to calculate the Mean Life and use the Units drop-down list to make sure the results will be returned in hours. The mean life of the lamp is estimated to be about 11,145 hours, as shown next.
You continue to use the QCP to solve for the remaining reliability metrics:
You select BX% Life to calculate the time by which 10% of the lamps will be failed. The result is about 1,913 hours. You select Reliability to obtain the reliability at 200 hours of operation. The result is 0.993465, or 99.3465%. To solve for the upper bound on the number of LED lamps (out of 1,000) that would fail after 200 hours, follow these steps:
Select Prob. of Failure and enter 200 for the mission end time to calculate the probability of failure at 200 hours. The probability of failure at 200 hours is estimated to be 0.006535 or 0.65%. Now multiply this by the total number of lamps (1,000). The number of lamps that will fail is 6.5, or approximately 7.
To estimate the warranty time during which no more than 2% of the lamps will fail, you select to calculate the Reliable Life, and then enter 0.98 for the Required Reliability input. The result is about 499 hours. Alternatively, the same result could have been obtained by calculating the B2 life.
39
DesignofReliabilityTests
Weibull++ includes a number of test design tools that provide ways to design reliability tests and evaluate/compare proposed test designs. In this chapter, you will work with the three tools available in the Test Design folio, starting with the Reliability Demonstration Test tool, which you will use to design a zero-failure test intended to demonstrate a specified reliability for bulb A. Afterwards, youll examine ways to compare the reliability of bulb A and a less expensive bulb B. To do this, youll use the Difference Detection Matrix to determine what test times are required to demonstrate a difference in reliability between the two bulbs. Then youll use the Expected Failure Time Plot to compare the failure times that are expected from a design with a specified reliability to the actual failure times that are observed during a test.
4
Zerofailure demonstration test Detecting differencesin reliability Predicting expectedfailure times
4.1CombiningtheTestDataforBulbA
Before you can use the test design folio, you need to estimate the life distribution followed by bulb A. You already have data from two tests for the bulb, and you established previously that there is no significant difference between the two data sets (see Chapter 2). You decide to combine the data sets and analyze them in a separate standard folio (Bulb A - Combined Data in the sample project). The analysis shows that the failure behavior of bulb A follows a 2-parameter Weibull distribution with beta = 4.219615. Using the Quick Calculation Pad, you also see that the estimated B10 life of the bulb is about 581 hours.
41
4DesignofReliabilityTests
4.2ReliabilityDemonstrationTestDesign
Based on your experience with analyses for bulb A, which is currently being used in the projector, you are asked to design a demonstration/validation test to show that the bulb has the required B10 life. More specifically, your objective is to demonstrate that bulb A has a B10 life of at least 500 hours with 80% confidence. You are allocated 10 bulbs for the test and decide on a zero-failure reliability demonstration test. In other words, you decide to design a test that uses the available sample size of 10 and will demonstrate the target metric if no failures occur before a certain time (i.e., you design a test where the number of allowable failures is zero).
ZeroFailure Tests: A zerofailure test is appropriate in this case because you only want to know,
withagivenconfidencelevel,whetherthelifemetricisgreaterthanthespecifiedrequirement.This sortoftestprovidesaquickandefficientwayofaccomplishingthis.Note,however,thatzerofailure tests generally cannot be used to determine the actual value of a products life metric. (In some special cases, the 1parameter Weibull or 1parameter exponential distribution can be suitable for analyzingadatasetcontainingfewornofailures.)
Objectives
Determine the test time for a zero-failure test that would demonstrate the target metric with a sample size of 10. Compare the test times that would be required given different sample sizes and different numbers of allowable failures.
Solution
To create a new test design folio, you choose Insert > Tools > Test Design.
In the Test Design Assistant that appears, you select Reliability Demonstration Test Design. Since you are able to make reasonable assumptions about the life distribution that describes bulb As failure rate behavior, you select the Parametric Binomial test design method on the control panel of the folio (ZeroFailure Test - Bulb A in the sample project). To specify the metric that you wish to demonstrate, you choose Reliability value at a specific time from the Metric drop-down list on the RDT sheet. Then you enter 90 for the reliability and 80 for the confidence level. (The B10 life is equivalent to the time at which reliability reaches 90%.)
42
4.2ReliabilityDemonstrationTestDesign
Next, you click the Get Distribution from Folio icon and select the standard folio that analyzed the combined data sets for bulb A (Bulb A - Combined Data in the sample project).
This copies the life distribution from the standard folio into the Distribution and With this Beta fields of the tool, and it calculates the time at which bulb A would reach the specified reliability with the specified confidence level (i.e., the B10 life). The calculated time is about 581 hours, and it appears in the At this time field. You check to make sure the B10 life is equal to or greater than the B10 life you wish to demonstrate (i.e., 500 hours). Since it is greater, you know a demonstration test is possible, and so you continue planning the test by changing the value in the At this time field to 500. In the Solve for this value area, you select to solve for the Required test time for a test with a sample size of 10. To specify that this will be a zero-failure test, you enter 0 for the maximum number of failures. Finally, you click the Calculate icon. The inputs and results appear as shown next.
Based on this calculated result, you need to test the 10 bulbs for at least 553 hours each. If no failures occur during that time, you will have demonstrated that the bulb has a B10 life of at least 500 hours with 80% confidence (i.e., if no failures occur before about 553 hours, then there is an 80% probability that the B10 life is greater than 500 hours).
You would also like to explore the possibility of using a different sample size and examine the sample sizes effect on test time. This may allow you to find a better combination of sample size and test time. So you click the Create Table of Results icon on the control panel.
A sheet called Table of RDT is added to the folio. You select to solve for the Test time for given sample size and specify that you want to see the required test times for sample sizes that range from 5 to 20, using an increment of 1. Then, because you want to view test times for zero-failure and 1-failure tests, you specify that you want to consider numbers of failures that range from 0 to 1. Finally, you click Calculate and see the results shown next.
This table uses the target metric and life distribution information you provided earlier to calculate the test times needed under many different scenarios. For example, the first highlighted cell shows that you could reduce the sample size to 7 and test to 602 hours instead of 553 (with zero failures). The second highlighted cell shows that if you were to stay with your current sample size of 10, you could demonstrate the target metric even if 1 failure occurred. However, the required test time in that case would increase to 649 hours.
44
4.3DifferenceDetectionMatrix
To more easily see how sample size, test time and number of failures are related, you generate a plot of this information by clicking the Plot icon. The plot is shown next.
The Reliability Demonstration Test tool is used to plan a test that is intended to demonstrate that a products reliability exceeds a specified target. Sometimes, however, tests will be performed just for comparative purposes (e.g., to show that one products B10 life is greater than another products). The Difference Detection Matrix calculates how much test time is required before it is possible to detect a statistically significant difference in a reliability metric (e.g., mean life) of two product designs by analyzing the data from a reliability life test. The purchasing group has located a less expensive bulb B from another manufacturer. Based on the extremely attractive pricing of bulb B, you are asked to evaluate and propose plans to compare the B10 life of bulb B to the life of bulb A. You are also informed that you only have 5 units of bulb B available for testing. The goal of the comparison is to determine if there is a significant difference between the B10 lives of the bulbs at a specified confidence level. Because of the smaller sample size, you decide to use a confidence level of 70%. You realize that you could simply test bulb B and compare the results to the results obtained from testing bulb A. However, given that only 5 units of bulb B are available for testing, you decide to first explore the capability of a test (and the available sample sizes) to show any difference. In other words, you will be
evaluating various test durations to determine which durations are capable of demonstrating a difference in B10 life between the two bulbs.
Objective
Determine how long you must test the 5 units of bulb B to be able to detect a difference in the B10 life of the two designs.
Solution
You start by choosing Insert > Tools > Test Design and then selecting Difference Detection Matrix in the Test Design Assistant. The new matrix (Difference Detection - Bulb A vs. Bulb B) will allow you to see what test times will be required to show a difference between the bulbs, given a range of different B10 life values. On the control panel of the new folio, you specify that you will be comparing the B10 life by choosing BX% Life and entering 10 in the Metric to Compare area. You enter 70 for the confidence level and choose Hour (Hr) for the time units, as shown next.
You provide information about the sample size and life distribution of bulb A in the Design 1 area, and you provide this information for bulb B in the Design 2 area. The results from combining and analyzing the 25 data points from the in-house test and the suppliers data (Bulb A - Combined Data in the sample project) show that bulb A has a 2P-Weibull distribution with a beta of about 4.2. Since you will be testing the assumption that bulb A and bulb B have a similar reliability, you assume the same life distribution information for both products.
46
The fields in the Reliability Metric Setup area define the range of reliability metric values in which you want to see the results. Given that bulb A has a B10 life of about 600 hours, a reasonable range may be from 0 to 1,000, so you enter 1000 in the Max Metric Time field. To specify that you would like to see the results spread out over 50-hour increments, you enter 50 in the Metric Increment field.
The inputs in the Test Time Matrix Setup area define the test durations that you want to explore. After reviewing the analyses of bulb A, you set the range to have a maximum of 2,000 hours of test time and decide to have the software give you results at 9 other intervals. You do so by entering 10 in the Number of Test Times field, and then you clear the Calculate Test Times check box and enter the specific test times shown next.
With these settings, the software will evaluate test plans with ten different durations: 2,000 hours, 1,500 hours, and so forth.
47
With all the inputs provided, you click the Calculate icon to generate the matrix. Part of the matrix is shown and explained next.
Since the B10 life of Design 1 (bulb A) was estimated to be 581 hours, you can focus on either the 550 column or the 600 column. You choose to focus on the 550 column, as shown above. (You decide not to change to a 10-hour increment to view the 581 hours column, since that would make the matrix quite large.) You do not know what the life of Design 2 is; thus, you are interested in the range of life values that you specified on the control panel. The legend assigns an index number to each of the 10 test durations you specified in the Test Time Matrix Setup area. In this case, the index number represents the test time that would be needed to detect a difference in B10 life with 70% confidence. For example, suppose Design 2 has a B10 life of 350 hours. The index number in row 350 is 5, which corresponds to a test time of 600 hours. According to this result, if Design 2 has a B10 life that is equal to or less than 350 hours (and failure rate behavior that is similar to Design 1; in other words, behavior that follows a Weibull distribution with beta = 4.2), then if you tested 5 bulbs from Design 2 using the indicated test time, you would detect a difference between the designs 70% of the time (i.e., the difference
48
4.4ExpectedFailureTimePlot
would be demonstrated with 70% confidence). You click the cell to view the required test time and the confidence bounds that were used to determine that the test time demonstrate a difference.
Similarly, if Design 2 has a B10 life of 700 hours, then the index number is 4, which means that testing 5 of the bulbs for 800 hours would demonstrate a difference in B10 life with 70% confidence. Furthermore, if the B10 life of Design 2 is greater than 350 hours but less than 700 hours, then the index assigned is 0, which means that even a test of 2,000 hours is incapable of demonstrating, at the specified confidence level, a statistically significant difference between the two designs. Tip: The granularity of the results is based on the test times specified in the Test Time Matrix Setup area. If needed, you can iteratively adjust the test times and recreate the matrix in order to arrive at an optimal granularity.
Before performing a reliability life test, you can use the Expected Failure Time Plot to provide an overall expected timeline and sequence of events for the test. This in turn can be used during the actual test to monitor its progress for any early warnings that indicate a deviation from the original assumptions or expectations. After a discussion with management, you are asked to go ahead and run the life test on the five bulbs.
49
Objectives
Use the Expected Failure Time Plot to determine what range of failure times you can expect to observe on each unit of bulb B during this test. Assume that bulb A and bulb B have the same reliability. As the units of bulb B fail during the test, compare the failure times to the expected times. If the bulbs appear to differ in reliability, determine which has a greater B10 life.
Solution
To create the plot, you choose Insert > Tools > Test Design and select Expected Failure Time Plot in the Test Design Assistant. In the Plot Setup area on the control panel, you enter 5 for the sample size and 70 to specify that you wish to see 70% two-sided confidence bounds for each expected failure time. You enter 1 for the acceleration factor because the units will be tested under normal stress conditions. You need to associate the plot with the combined data for bulb A. To do this, you click the Get Failure Model from Folio icon on the control panel and select the calculated data sheet that contains the combined data from the manufacturer (Bulb A - Combined Data in the sample project).
The information in the Expected Failure Model area of the control panel is now updated with the distribution and parameters that were calculated from the selected data sheet. You click the Plot icon to
50
display the ranges of expected failure times (Expected Failure Times - Bulb B in the sample project), and then you adjust the x-axis range so the maximum value is 1,500. The plot appears as shown next.
This plot uses the failure number on the y-axis (1 = the first failure, and so on) and time on the x-axis. Based on the plot, you expect the first failure to occur sometime between 440 and 787 hours of testing, the
51
second between 635 and 928 hours, and so forth. You view the actual time bounds by pointing to the first or last tick mark on a line, as shown next for the first failure.
The Expected Failure Time Plot: In general, this plot tells you that, based on the assumed
distribution, you can expect the observed failure times to be within the ranges shown. If any sequentialobservedtimeisoutsidethisrange,thenthedistribution(andthusthereliability)ofthe observed units should no longer be assumed. (The plots results are only valid, however, for the confidencelevelchosen.)Moreover,ifanyoftheobservedfailuretimesisearlierthanitspredicted interval,thenyoucaninferthatthereliabilitywillbelessthanwhatwasassumed.Similarly,ifanyof theactualfailuretimesisgreaterthanitspredictedinterval,thenyoucaninferthatthereliabilityis greaterthanwhatwasassumed(atthespecifiedconfidencelevel). As the test progresses, you observe the first failure at 300 hours, which is earlier than the associated starting interval. You conclude that, at the specified confidence level, bulb Bs reliability is less than what was assumed. In order to estimate the bulbs B10 life, you continue the test to collect more data. The next 2 failures are observed at 440 and 495 hours. As these failures occur, you display them on the plot by
52
clicking inside the Actual Failures field on the control panel and entering them into the table that appears. After recording these 3 failures, the table appears as shown next.
The plot appears as shown next. With three failure times observed, you end the test at 500 hours and calculate the data.
53
You create a standard folio for the bulb B data (Bulb B - In-House Data in the sample project) and use the QCP to calculate the B10 life.
The estimated B10 life for bulb B is about 280 hours, which is clearly less than the 575-hour B10 life that was calculated for bulb A using the combined data. You decide to perform further analysis to determine whether this is a statistically significant difference. To do this, you choose Insert > Reports and Plots > Overlay Plot and select to compare the data sheets for the two bulbs. On the control panel of the overlay plot, you click Confidence Bounds and then select to view the 70% two-sided bounds on reliability. Finally, you click Target Reliability on the control panel and select to display the target reliability value of 90%. This will mark the time at which reliability equals
54
90%, which is equivalent to the B10 life. The plot (Bulb A vs. Bulb B in the sample project) appears as shown next (with annotations added via RS Draw to make the plot easier to interpret).
The horizontal line marks the target metric, and you can see that bulb Bs B10 life is lower than that of bulb A, and that the confidence bounds on the B10 life for both products do not overlap. Thus, you conclude that bulb B indeed has a lower B10 life. You show the plot to the design team and recommend that they use bulb A in the projector. Note: In the difference detection matrix you created in Section 4.3, you can see that if the B10 life of bulb B is 300 hours and the B10 life of bulb A is approximately 550 hours, then a 500hour test with a sample size of 5 wouldbesufficienttodetectadifference(ata70%confidencelevel)betweenthedesigns.Noticethattheresults inthissectionareconsistentwiththepriorresultsandthatadifferencewasdetectedonthis500hourtest.
55
WarrantyAnalysis
Development testing allows you to uncover and correct reliability problems before a product is deployed; however, there are instances when a problem is not discovered until the product is in the customers hands. Data obtained from field failures can provide valuable information about how a product actually performs in the real world. In this chapter, you will extract life data from warranty returns records, and then compare the results obtained from the field data to the results obtained from an in-house reliability test.
Warrantydata analysis Nevadachart format Forecasting productreturns
5.1WarrantyDataAnalysis
Based on your prior analyses, you recommended bulb A because of its higher reliability; however, bulb B is lower-priced, and so a decision was reached by management to use bulb B for production. The projectors with bulb B were then released to market with a 12-month warranty. Soon after the product introduction, due to excessive returns (as you predicted), management decided to discontinue the use of bulb B and return to bulb A, which you originally recommended. Specifically, bulb B was used in production for twelve months. After that, any projectors released in the 13th month and beyond were modified to use bulb A. Six months after this change, you are asked to forecast the warranty returns due to bulb failures for the next six months. The returns data are available and shown in the
57
5WarrantyAnalysis
following table. The data set is organized in Nevada chart format. The data includes all 18 months of service (12 with bulb B and 6 with bulb A).
In this format, the columns represent the warranty returns periods and the rows represent the sales periods. This chart indicates, for example, that a total of 18 units were returned during the 3rd month (column 3), and that 13 of those units were sold in the first month (row 1) while 5 units were sold in the 2nd month (row 2). The rest of the chart can be read in a similar manner.
Objectives
Assuming an average sales of 1,000 projectors per month, estimate the number of warranty returns due to bulb failures that will occur in the next six months (i.e., months 19 to 24). As a side analysis, verify whether an assumption of a usage rate of 50 hours/month per projector holds true for bulb B. Do this by comparing the record of warranty failures to the data from the inhouse test of the bulb. See page 49 in Chapter 4.
Solution
To create a warranty analysis folio, you choose Insert > Folios > Warranty.
When prompted to specify the data format, you select I want to enter data in the Nevada chart format, and then click Next. Currently, there are 18 sales periods (i.e., the projectors with bulb B were sold in the first 12 months and the projectors with bulb A were sold in the following 6 months) and 18 returns periods. You are asked to
58
forecast the returns for the next 6 months, so you include the forecasted future sales for the next 6 months. Note that because all new projectors have bulb A in them, the future sales will be of projectors with this type of bulb. You set up your Nevada chart with the following settings:
Note: In the Warranty Folio Setup window, you could select to label the periods in terms of months instead of
numbers;however,thesetupwillrequireyoutospecifythecalendaryear.Thisexampledoesnotanalyzeseveral years of data, so we will simplify the data entry process by labeling the periods in terms of numbers instead of months.
Once the folio is created (Warranty - Bulbs in sample project), you click the Change Units icon on the control panel.
You then specify that the time units of the sales and returns periods will be in terms of months.
59
5WarrantyAnalysis
Next, you enter data in the Sales data sheet. You use the assumption of a constant rate of sales of 1,000 projectors per month, and use the subset ID column to specify the type of bulb that was used in the projectors that were sold that month. Note:SubsetIDsallowyoutoseparatethefulldataintodifferenthomogeneousdatasets.Inthisexample,there are two products, bulb A and bulb B. Separating the data by product will allow you to fit different models (distributionandparameters)toeachone. The resulting Sales data sheet is shown next.
You then enter the data from the Nevada chart on page 58 to the Returns data sheet of the folio.
60
On the control panel, you select the Use Subsets check box, as shown next.
This setting tells the software to look at the information in the subset ID column and separately analyze the data based on their subset IDs. In addition, it allows you to use the drop-down list to switch between subset IDs and alter the analysis settings for each one, if appropriate. As a general rule, however, the recommendation is to use the same analysis settings for all subsets. You select the following analysis settings for each subset ID:

2P-Weibull Maximum Likelihood Estimation (MLE) Standard Ranking Method (SRM) Median Ranks (MED) Fisher Matrix Confidence Bounds (FM)
Next, you select the Suspend After check box and enter 12 periods. Note: In many cases, data after the warranty period are incomplete or unreliable. The Suspend After setting
allowsyoutoignoreanyfailuretimesafterthespecifiedperiodandtreatthemassuspensions(i.e.,allweknowis thattheunitshadnotfailedbythattime).
Once the settings are finalized, you analyze the data sheet by clicking the Calculate icon on the control panel.
61
5WarrantyAnalysis
The Analysis Summary area of the control panel will display two sets of parameters (because the data sheet contains two subsets IDs). You view the parameters of each set by choosing the subset ID from the drop-down list, as shown next. The following picture shows the 2P-Weibull parameters of bulb B.
The following picture shows the 2P-Weibull parameters of bulb A.
With the analysis completed, the next step is to generate the forecast. You do so by clicking the Generate Forecast icon on the control panel.
62
In the setup window, you set the forecast to start on the 19th month and show 6 periods, as shown next.
The forecast is then generated within the folio and displayed in the data sheet called Forecast. On the control panel of the Forecast sheet, you select the Show Subset ID check box in order to identify which type of bulb the expected failure will be. You also select the Use Warranty Length check box, enter 12 periods and click Update. Note: By default, the software assumes an infinite warranty period. The Use Warranty Length check box allows
you to limit the length of the warranty period. In this example, the warranty length is 12, which means that any units that fail after 12 months of operation are not counted in the returns as they are assumed to be out of warranty.
63
5WarrantyAnalysis
The following Forecast sheet shows the results of your analysis, with the Show Subset ID and Use Warranty Length options selected.
The columns represent the warranty returns periods and the rows represent the sales periods. The results indicate, for example, that you would expect to see a total of 616 bulb failures in the 19th month (column 19), and that 32 of those failures will be from bulb A (as shown in rows 13 to 15), while the remaining failures will be from bulb B. The rest of the forecast can be read in a similar manner. As a side analysis, you want check whether the assumption of a usage rate of 50 hours/month is valid. The data from the in-house test of bulb B are in terms of hours, so you first convert the data for bulb B according to the hour-to-month ratio (by dividing the failure and suspension times by 50). You create a new Weibull++ standard folio and manually enter the converted data. You analyze the data sheet using the same analysis settings that were used on the in-house test data (i.e., 2P-Weibull and RRX).
64
The following data sheet (Bulb B - In-House Data - Months in sample project) shows the result of the analysis (for reference, the subset ID column shows the failure times in terms of hours).
You can now compare the results from the warranty analysis to the results from the in-house test and verify the assumption of the usage rate. One way to do this is to display the contour plot of both data sets and analyses on the same plot. To do this, you create an overlay plot by choosing Insert > Reports and Plots > Overlay Plot.
When prompted to select which data sets to plot, you select the converted data set from the in-house test and the data set from the warranty analysis of bulb B (Bulb B - In-House Data - Months and Warranty - Bulbs in the sample project). On the control panel of the plot sheet, you switch the plot type to a Contour Plot. When prompted to specify the contour lines, you select the 2nd Level, 90% and the 5th Level 75% check boxes. The following overlay plot (Bulb B - Warranty vs. In-House in sample project) shows that all the contours overlap; therefore, the two data sets do not show a statistically significant difference in the life,
65
5WarrantyAnalysis
and the 50 hour/month assumption is valid. (This assumes that the only factor is the hour-to-month ratio and that there is no differentiation between the test and use conditions.)
66
TargetReliability Estimation
Deciding on a reliability goal/target involves trade-offs. This is because higher reliability typically correlates with higher production costs, lower warranty costs and higher market share. With Weibull++'s Target Reliability tool, you can generate plots that help you visualize and estimate a target reliability that will minimize cost, maximize profit and/or maximize your return on an investment in improving the products reliability. In this chapter, you will use cost factors to estimate the target reliability for a product and calculate the return on an investment intended to reach that reliability.
Estimatingtarget reliability Calculatingthe returnonan investmentin reliability
6.1ReliabilityandReturnonInvestment
Based on their benchmarking and market research, the marketing group believes the company will become the market share leader if it offers a highly reliable projector coupled with a one-year warranty period. Specifically, their research shows that if the company maintained the current sale price of $1,500 per unit, used a one-year (600 hours) warranty and managed to provide a projector with a 99% reliability over the warranty period, its current 20% market share would increase to 45%. On the other hand, a highly unreliable product (more than 50% returns under warranty) would virtually eliminate the companys market share (i.e., reduce it to about 1%). The projector bulb is covered under the warranty and is the projectors primary cause of failure. Based on the prior in-house analyses of bulb A and bulb B (see Chapters 2 and 4), if the company were to use bulb A, it is expected that 12% of the projectors would fail during the warranty period. Furthermore, if the company decided to use the lower priced bulb, bulb B, it is expected that 67% of the projectors would fail under warranty. Moreover, after looking at additional manufacturers, you learn that more reliable bulbs
67
6TargetReliabilityEstimation
are available, but at a higher cost. You identify one manufacturer that offers a highly reliable bulb (1% failures during the warranty period). The cost of producing a projector without a bulb is $400. The additional cost for the bulb varies depending on which bulb is used.

The least expensive bulb (bulb B): $50 each. The moderately reliable bulb (bulb A): $175 each. The bulb with the highest reliability: $450 each.
Finally, each warranty claim costs $200 to process, in addition to the replacement cost of the bulb.
Objectives

Estimate the target reliability for the projector bulb based on the one-year warranty period. Assume that a redesign is required to achieve the target reliability and that such a redesign effort is estimated to require an initial investment of $10 million. Estimate the return on this investment.
Solution
To open the Target Reliability tool (Bulb B - Target Reliability in sample project), you choose Insert > Tools > Target Reliability.
In the Target Estimation Inputs area, you use the information given above to describe the following three scenarios, all of which assume a one-year warranty:

Best Case: Purchase the most expensive, most reliable bulb. Mostly Likely: Purchase bulb A, the modestly reliable bulb.
68
Worst Case: Purchase bulb B, the least expensive and least reliable bulb.
Here is a more detailed breakdown of the inputs:
Expected failures/returns per period (as % of sales): With the most expensive bulb, 1% of the projectors would be returned during the warranty period. With bulb A, 12% of the projectors would be returned. With bulb B, 67% would be returned. % of market share you expect to capture: Your market share in the projector market is 20% with bulb A. This could be increased to 45% with the most reliable bulb or reduced to 1% with bulb B. Average unit sales price: The sales price of the projector will be fixed at $1,500, regardless of the type of bulb used. Average cost per unit to produce: Each projector costs $400 to produce, plus $450 for the most expensive bulb, $175 for bulb A and $50 for bulb B. Other costs per failure: Additional costs are assumed to be fixed at $200, regardless of which bulb is used.
As reliability increases, your production costs will increase, but your unreliability cost will decrease. The Cost vs. Reliability plot allows you to visualize the relationship between reliability and cost, as well as estimate the reliability that minimizes total cost (i.e., production cost + cost due to unreliability). After
69
6TargetReliabilityEstimation
you click the Plot icon on the control panel, the Cost vs. Reliability plot appears as shown next (with annotations added via RS Draw to make the plot easier to interpret).
As the legend in this plot shows, the reliability that is estimated to minimize cost is 96.7% at the end of the warranty period. Next, you select to view the R3OI vs. Reliability plot so you can estimate the reliability target that has the highest ROI using ReliaSofts Reliability Return on Investment (R3OI) equation. This plot requires you to enter the maximum market potential for the product (in number of units sold). The total market share is estimated to be 120,000 units over the life of the product, so you enter this value in the Max. Market Potential (Units) field on the control panel and click the Redraw Plot icon.
70
The plot appears as shown next.
According to the plot legend, if your goal is to maximize the return on investment, then your target reliability should be about 96.6% at the end of the warranty period (which is obviously very close to the 96.7% reliability needed to minimize cost). With this reliability, you can expect 3.4% of units sold to be returned under warranty. Next, you want to determine the return on an investment to reach the target reliability, assuming that an initial investment of $10 million would be required. So you enter 10,000,000 in the Initial Investment field and redraw the plot. Then, to see the actual ROI, point to the vertical green line to highlight it.
You conclude that the return rate on a $10 million reliability investment would be about 276.7%.
71
StressStrength Comparison
Generally, the reliability of a product is calculated based on its ability to perform without failure for a specified period of time. In some cases, however, you may need to obtain the reliability of a product based on its ability to withstand an applied stress without failure, rather than the length of time it continues to operate. An example is a wooden beam that may buckle under a load that is over a certain weight limit. In this chapter, you will estimate the reliability of a product based on the strength of the material and the stress it will experience.
StressStrength tool Reliability Demonstration Test(RDT)tool
7.1StressStrengthAnalysis
A different design group in the company needs your assistance in selecting and qualifying a steel component for a new product. Their main concern is the ability of the component to withstand the tensile stresses in its intended application. The estimated median tensile stress is 75 Mpa. The worst case load (defined as the 99th percentile case) is estimated to be 150 Mpa. The reliability requirement for the component is a probability of failure of 1 in 100,000 due to tensile loads (or a reliability of 99.999%). The group had previously conducted a test on six specimens. They send you the following data set, which shows the stress values at which each of the specimens failed: 212, 236, 248, 254, 271 and 287 Mpa. They have 10 more specimens available for testing if the current information is not sufficient to demonstrate the reliability target.
73
7StressStrengthComparison
Objectives
Determine whether it is possible to demonstrate the required reliability with the information provided. If not, design a test for 10 specimens that will demonstrate that the component meets the required reliability if all specimens pass the test.
Solution
The first step is to obtain a distribution for the tensile stress that the component will experience in the field. To do this, you create a Weibull++ standard folio by choosing Insert > Folios > Weibull++ Standard Folio.
When prompted to specify the data type, you select Free-form (Probit) data. Note: Because the data set you are analyzing is in terms of stress rather than time, you have two choices. You can set the units for the data sheet to Hours and then remember throughout the analysis that any inputs and results given in Hours really means megapascals (Mpa). Alternatively, if you have administrative permissions in the Synthesis repository (File > Manage Repository > Manage Units), you can define a new unit for Mpa that will then be available in the Units dropdown list for any analysis performed within the database. For the purposesofthisanalysis,itisnotnecessaryforthesoftwaretobeabletoconvertMpatosomeotherunitsowe willusethesimplerapproachofinterpretingHoursasMpa. Once the folio is created (Steel Component Data in the sample project), you double-click the name of the data sheet (located at the bottom of the window) and rename Data1 to Stress. You then enter the estimates for the tensile stress in the data sheet. On the control panel, you choose the following analysis settings:

2P-Weibull Rank Regression on X (RRX) Standard Ranking Method (SRM) Median Ranks (MED)
To analyze the data set, you click the Calculate icon on the control panel.
74
The following data sheet displays the result of your analysis. It shows that the stress distribution is a 2PWeibull distribution with beta = 2.732021 and eta = 85.767713 Mpa.
Tip:Becauseyouhavetwoprobabilitypoints(i.e.,theunreliabilityisestimatedtobe50%at75Mpaand99%at
150 Mpa), youcouldalsousethe QuickParameterEstimator(QPE)toobtaintheparametersofthedistribution. Toaccesstheutility,chooseHome>Tools>QuickParameterEstimator(W++).
IntheQPE,selecttheTwoUnreliabilityPointsoptionandthe2PWeibulldistribution.Enterthestressvaluesand unreliabilityvalues,andclickCalculate.Youwillobtainthesameestimatesfortheparameters.
The next step is to obtain the strength distribution. You create a second data sheet in the folio by rightclicking the data sheet tab area (the area at the bottom of the window that shows the name of the data sheets) and choosing Insert Data Sheet on the shortcut menu. This time, you select Times-to-Failure data as the format and clear all the other options.
75
You rename the second data sheet to Strength, and then enter the failure times from the test (i.e., the stress value at which each specimen failed). On the control panel, you select the same analysis settings, and then click Calculate to analyze the data. The following data sheet displays the result of your analysis. It shows that the strength distribution is a 2PWeibull with beta = 10.056225 and eta = 262.950892 Mpa.
The next step is to compare the stress and strength distributions. You open the Stress-Strength tool by choosing Insert > Tools > Stress-Strength.
You select the Stress and Strength data sheets, and click OK.
76
The Stress-Strength tool shows the following plot (Stress-Strength - Steel Component in the sample project).
In addition, the result on the control panel tells you that the reliability of the component is estimated to be 99.980787%. This value is close to the target reliability, but not close enough. You know that the design group has 10 more specimens that they could use to test against the tensile loads. You want to use a zero-failure test (also called reliability demonstration test or RDT), which will demonstrate that the component meets the required reliability if all specimens pass the test. To design the test, first you need to determine how strong the component needs to be in order to meet a reliability target of at least 99.999%. This means that you need to determine the parameters for the strength distribution that, when compared against the stress distribution in the Stress-Strength tool, would result in the required reliability. To find out, you click the Target Reliability Parameter Estimator icon on the control panel. 75
77
You select to estimate a new eta parameter for the strength distribution, and then you enter the target reliability, which is 99.999%. When you click Calculate, the result tells you that in order to meet the reliability target, the value of eta needs to be 356.417171, as shown next.
Based on this result, you can now design a zero-failure test plan. To do this, you click Update to transfer the new eta value to the Stress-Strength tool. You can see that in the control panel, the parameters of the strength data set are now labeled (Altered). You click the Reliability Demonstration Test icon of the strength data set, as shown next.
This automatically creates a new RDT folio (Test Design Folio - Steel Component in the sample project) and populates the test plan with the MTTF of the strength distribution that you want to be able to demonstrate. The software automatically populated all of the Parametric Binomial test plan inputs,
78
including the companys standard 90% confidence level for demonstration tests, so you just need to choose the following settings in the Solve for this value area:

Value: Required test time Sample Size: 10 Maximum Number of Failures: 0
In this analysis, the required test time is the tensile stress that the specimens must experience without failure. You compute the results and it shows a test time (i.e., tensile stress) of 307.990114 Mpa, as displayed next.
You inform the group of your findings and show them the test plan you created. You explain that if they test 10 specimens at a tensile stress of 307.990114 Mpa and there are no failures, then this will demonstrate that the component meets the required reliability with a 90% probability.
79
CompetingFailureModes Analysis
Competing failure modes analysis is a method of analyzing the reliability of a product that has more than one cause of failure. The analysis uses the following assumptions: a) the failure modes are independent, meaning that the occurrence of one mode does not affect the probability of occurrence of the other modes, b) the system will fail if any of the modes occur (i.e., series configuration) and c) the failure rate behavior for each failure mode is known and can be described with a life distribution and parameters. In this chapter, you will work with a product that experiences multiple failure modes and explore two ways to perform the analysis.
Competingfailure modesanalysis Batchautorun Reliabilityblock diagrams
8.1CompetingFailureModes
Based on your prior work, your reputation for reliability analysis work has reached other divisions within the company. Another division requests your assistance on a classified project, but because you do not have the required security clearance to access classified information, you have no specific knowledge about the product. The only information that can be shared with you is that the product has three failure modes (A1, A2 and A3), and that it would fail if any one failure mode occurs.
81
8CompetingFailureModesAnalysis
Thirty identical units were tested. The failure times (in hours) for the test units, along with the failure mode responsible for the failure, are given next.
Objectives
Use the built-in Competing Failure Modes (CFM) analysis method in a Weibull++ standard folio to perform the analysis and estimate the median time to failure of the product (i.e., B50 life). Perform the analysis again using a reliability block diagram (RBD), and estimate the B50 life again to show that the results are the same. Determine whether the product can meet a required B10 life of 350 hours. If it cannot be achieved, determine which failure mode will need to be addressed in order to meet the requirement.
82
Solution
First, you create a new Weibull++ standard folio by choosing Insert > Folios > Weibull++ Standard Folio.
When prompted to specify the data type, you select Times-to-failure data and clear all the other options because your data set only contains complete data. You also use the Units drop-down list to indicate that the time values in the data sheet will be entered in hours. Once the folio is created (Competing Failure Modes in the sample project), you enter the failure times and the associated subset IDs that identify the failure modes. On the control panel, you choose Competing Failure Modes > CFM-Weibull. The Settings area of the control panel provides a quick summary of the analysis settings that will be used to fit the distribution. The current settings on the Main page of the control panel are as follows:

Tip:TheseoptionscanbesetfromtheAnalysispageofthecontrolpanelorbyclickingthebluetextlabelsinthe SettingsareaontheMainpage. Next, you analyze the data (fit the selected distribution to each failure mode) by clicking the Calculate icon on the control panel.
When you select to use CFM analysis, the software prompts you to define up to four failure modes for the analysis. Each mode can be associated with one or more of the subset IDs of the data sheet. You move
83
each subset ID (on the left) under a different mode (on the right). When you are finished, the window looks like the picture shown next.
You click OK to begin the analysis. When the data set has been analyzed, the Analysis Summary area of the control panel displays the estimated parameters of the failure distribution for one mode at a time. You view the parameters of each mode by choosing the mode from the drop-down list, as shown next.
Next, you open the QCP by clicking the icon on the control panel.
To calculate the B50 life, you select BX% Life and use the Units drop-down list to make sure that the results will be returned in hours. You enter 50 for the BX% Life At input and click Calculate.
84
The result will appear, as shown next.
A B50 life of approximately 561 hours means that 50% of the population will have failed by 561 hours of operation. Because some members of the design team are not familiar with the built-in CFM analysis method available in Weibull++ standard folios, the next step is to perform the analysis again using the more familiar RBD approach to demonstrate that both methods return identical results when all of the settings and assumptions are the same. Before you create a diagram, you need to segregate the data sets of the failure modes so that each mode can be represented by a single block in the diagram. You click the Batch Auto Run icon on the Main page of the control panel.
When prompted to select the subset IDs, you select them all.
85
You click the Processing Preferences tab. In the Extraction Options area, you select the option shown next.
This option allows you to extract the data sets that are required for the CFM analysis. For example, when you extract the data points of mode A1, the Batch Auto Run utility will retain all the data points due to mode A1 as failures but mark all other data points due to modes A2 and A3 as suspensions. They are suspensions because the units in the sample would have continued to operate for some unknown amount of time if they had not been removed from the test when they failed due to those modes. In the Calculation Options area, you select to calculate the parameters of each data set based on the selected distribution, which is the 2P-Weibull for this analysis. When you click OK, the data sets are extracted to separate data sheets in the folio and automatically calculated. To create a diagram that allows you to analyze the extracted data sets in a series configuration, you choose Insert > Tools > Diagram.
Once the diagram is created (Competing Failure Modes - RBD in the sample project), you add a block by right-clicking the diagram and choosing Add Block on the shortcut menu. When prompted to select the data sheet of the failure mode that the block will represent, you select the data sheet for mode A1. You use the same approach to add the blocks that will represent failure modes A2 and A3.
86
Next, you connect the blocks in an appropriate configuration to describe the relationships between the failure modes. You add a connector by right-clicking the diagram sheet and choosing Connect Blocks on the shortcut menu. You click the source block, hold down the left mouse button and then drag a line from the source block to the destination block. When the crosshairs are located above the destination block, you release the mouse button to create a connector. The following diagram shows the modes connected in a series configuration. This indicates that the product will fail if any of the modes occurs.
To analyze the diagram and calculate the products overall reliability, you click the Calculate icon on the diagrams control panel. To estimate the B50 life of the product, you use the QCP. The result is about 561 hours, which is identical to the result you obtained when you performed the CFM-Weibull analysis in the Weibull++ standard folio. Tip: The RBD approach can also be used for more complex analyses that are not supported by the CFM analysis
method built into the Weibull++ standard folio. Specifically, you will need to use the RBD approach if: a) the analysis involves more than four failure modes, b) the failure modes are described by different life distribution and/orc)therelationshipsbetweenthemodesdonotfollowaseriesconfiguration.
You return to the CFM-Weibull analysis in the Weibull++ standard folio (Competing Failure Modes in the sample project), and use the QCP to calculate the B10 life of the product.
87
The required B10 life is 350 hours, but the result is about 129 hours, as shown next.
To visualize how each of the failure modes contribute to the overall reliability of the product, you click the Plot icon on the control panel.
The following probability plot shows the results (with annotations added via RS Draw to make the plot easier to interpret).
88
The plot tells you that mode A2 has a strong influence on the overall probability of failure of the product. You decide to perform a what-if analysis to see what the B10 life of the product would be if the failures due to mode A2 did not occur. To do this, you create a copy of the data sheet in the folio by right-clicking the data sheet tab area (the area at the bottom of the window that shows the name of the data sheets) and choosing Move/Copy Sheet on the shortcut menu. You select the data sheet that contains the data set, select the Create a Copy check box and click OK. Once the new data sheet is created, you rename it to Two Modes. You then click the Calculate icon on the control panel to redo the CFM analysis with only two modes. When prompted to select the subset IDs, you remove mode A2 from the list of modes, as shown next.
After performing the CFM analysis again for only modes A1 and A3, you use the QCP to calculate the B10 life. This time, the result shows 396 hours, which indicates that the product would be likely to meet the requirement if the designers were able to address the A2 failure mode (best case scenario).
89
ParametricRecurrent EventDataAnalysis
In life data analysis, it is assumed that the components being analyzed are non-repairable; that is, they are either discarded or replaced upon failure. However, for complex systems such as automobiles, computers, aircraft, etc., it is likely that the system will be repaired (not discarded) upon failure. Failures are recurring events in the life of a repairable system, and data from such a system are known as recurrent event data. Weibull++ includes a choice of two methods for analyzing recurrent event data: parametric and non-parametric analysis.
Recurrentevent dataanalysis ParametricRDA folios
In this chapter, you will perform parametric recurrent event data analysis in order to estimate the number of spare parts that will be needed to maintain a repairable system. Watchavideoofthisexampleat http://www.ReliaSoft.tv/weibull/startguide/weibull_qsg_9.html.
9.1ParametricRecurrentEventDataAnalysis(RDA)
You are performing consulting work for a customer who uses a piece of equipment that your company manufactures. The customer has 200 such systems, with each system using five bulbs. If any one of the bulbs in a given system fails, the system is considered failed. The customers maintenance department restores the system by replacing the failed bulb. When maintenance personnel go out to repair a system, the only information recorded is the use time of the system as indicated by the Hobbs meter on the equipment. The table on the next page shows the failure logs for 28 of those systems, all recorded between 0 to 1,500 hours of operation. Each log includes the operating hours of the system at the time when each failure occurred (indicated by an F) and the time when the observation period ended (indicated by an E).
91
9ParametricRecurrentEventDataAnalysis
Recurrent Event Data: In recurrent event data, the failure time that is recorded is not the
observation time when the failure occurred, but rather the age of the system when the failure occurred.Systemagemaybemeasuredintermsofthenumberofoperatinghoursaccumulated,or intermsofaccumulatedmileage,distance,etc.
1 467 792 1111 1177 1190 1333 1383 1500 2 F F F F F E 543 755 1310 1369 1388 1500 F F F F F E 3 826 978 1003 1019 1162 1500 F F F F F F F F E 4 326 571 1016 1025 1085 1104 1119 1420 1500 11 343 659 806 810 936 1111 1262 1500 18 421 721 1169 1212 1279 1489 1500 F F F F F F F F F E F F F F F F E 5 322 548 641 648 826 1087 1350 1482 1489 1500 12 589 877 912 1051 1218 1343 1500 19 564 1195 1370 1420 1500 F F F F F F F E 6 796 829 952 1053 1140 1375 1447 1500 F F F F F F F E 7 197 624 909 975 1012 1176 1370 1500
F F F F F F F E
F F F F F E
8 825 847 939 995 1209 1500
9 F F F F F F E 547 686 820 825 1276 1368 1500 16 370 531 547 1045 1052 1077 1283 1394 1452 1500 F F F F F E
10 659 913 916 1309 1319 1500
F F F F F F F E F F F F F F E
F F F F F F E
13 375 413 912 1032 1182 1287 1500 20 627 734 971 978 1395 1401 1500
F F F F F F E
14 714 794 977 1237 1279 1453 1500 21 141 545 804 967 1096 1154 1199 1500
F F F F F F E
15 449 582 898 1101 1154 1334 1500
F F F F F F F F F E
F F F F F F E
17 614 784 921 1100 1298 1475 1500
F F F F E
F F F F F F E
F F F F F F F E
92
F F F F F E
22 616 791 792 923 1413 1500
F F F F F F F E
23 657 717 762 909 1041 1277 1444 1500
F F F F F F F E
24 310 726 868 1009 1017 1292 1310 1500
F F F F F F E
25 415 559 847 982 986 1378 1500
F F F F F F F E
26 578 816 874 1082 1148 1434 1483 1500
F F F F F E
27 832 917 1024 1048 1463 1500
F F F F F F F F F E
28 156 505 717 764 855 903 1299 1371 1493 1500
Objectives

Estimate the number of bulbs that are expected to fail per system over the next 100 hours. Estimate the number of spare parts that the maintenance department will need to cover repairs for all 28 systems in the next 100 hours of operation.
Solution
For this data set, you know that you cannot use life data analysis (e.g., fit a Weibull distribution to the data set) because the failure times of each system are dependent on that systems history of repairs. In other words, all you know is that when a system failed, one of its five bulbs was replaced, but you do not know which one. In this case, from a system perspective, you are dealing with a repairable system (with the repair being the replacement of one of the five bulbs). Each time a system is repaired, the system is returned to service, but it is no longer in an as good-as-new condition nor as bad-as-old, but at some stage in between. To analyze the data and answer the questions of interest to the customer, you will perform parametric recurrent event data analysis using the general renewal process (GRP) model. First, you create a new Parametric RDA folio by choosing Insert > Folios > Parametric RDA.
Once the folio is created (Hobbs Readings Data in the sample project), you enter the data from the table on page 92 in the data sheet. After entering the data, the next step is to select the settings that will be used to fit the GRP model to the data. On the control panel, you select 3 parameters and the Type II virtual age model.
93
VirtualAge:Thegeneralrenewalprocessmodelusestheconceptofvirtualagetomathematically capture the effect of the repairs on the subsequent rate of occurrence of failure. The Type I model assumes that the repairs will remove some portion of the damage that has accumulated since the lastrepair,whiletheTypeIImodelassumesthattherepairswillremoveaportionofallthedamage thathasaccumulatedsincethesystemwasnew.
The setting for virtual age is also available on the Analysis page of the control panel, which can be accessed by clicking the Analysis button, as shown next.
On the Analysis page, you make sure that the Use Eta and RF parameterization check box is selected.
94
To fit the general renewal process (GRP) model to the data set, you click the Calculate icon on the Main page of the control panel.
Once the model is fitted, the Analysis Summary area of the control panel displays the parameters of the GRP model (beta, eta and RF) as well as the LK Value. The results are shown next.
Restoration Factor (RF): The RF parameter provides an indication of the effectiveness of the
repairs as reflected by the data set. An RF = 1 indicates that the system will be as good as new after the repair. This is known as a perfect repair. An RF value between 0 and 1 indicates that the systemwillbebetterthanoldbutworsethannewaftertherepair.Thisisknownasanimperfect repair. An RF = 0 indicates that the system will be as bad as old (i.e., no improvement) after the repair.Thisisknownasaminimalrepair. To solve for the number of bulbs that are expected to fail, you open the QCP by clicking the icon on the control panel.
To calculate the number of bulbs that are expected to fail per system over the next 100 hours, you select Cum. Number of Failures and use the Units drop-down list to make sure that the results will be returned in hours. You enter 1,500 for the Mission Start Time, and 100 for the Mission Additional Time.
95
The calculated result shows that 0.68 or about 1 bulb per system will fail in the next 100 hours, as shown next.
To estimate the number of spare parts that the customer would need to repair all 28 systems in the next 100 hours, you select Failures by System and enter 100 for the Mission Additional Time. This calculates the number of failures from the time when the observation period for the system ended (which is 1,500 hours as indicated by an E in the folio data sheet). You click Calculate to open the Results window that shows a detailed report on the number of failures per system. The result indicates that there will be a total of about 20 failures in the next 100 hours. Therefore, the customer will need 20 spare parts.
96
The Results window is shown next.
97
EventLogDataAnalysis
Event logs, or maintenance logs, capture information about a piece of equipment's failures and repairs, such as the date/time the equipment failed and the date/time the equipment was restored. This information is useful for helping companies achieve productivity goals by giving insight about the failure modes, frequency of outages, repair duration, uptime/downtime and availability of the equipment.
10

Eventlogfolios Overlayplots
In this chapter, you will learn how to use Weibull++s event log folio to convert maintenance logs into times-to-failure data, and then perform basic life data analysis. Watchavideoofthisexampleat http://www.ReliaSoft.tv/weibull/startguide/weibull_qsg_10.html.
10.1EventLogData
While working with the manufacturing division of your company, you are given the task of analyzing the failure behavior of a critical piece of equipment that operates for 24 hours a day, 7 days a week. The equipment has five different subsystems (subsystems 1, 2, 3, 4 and 5). Furthermore, you learn that:
When any one of the subsystems fails, the equipment stops operating until the failed subsystem is repaired (thus the remaining subsystems do not age and cannot fail during the repair action). A repair action involves replacing the failed subsystem with a new (similar) subsystem. An automated data logger automatically captures the equipment failure events and the subsystem responsible for the failure, as well as the date and time when the equipment was restored to operating condition. The equipment started operating on January 2, 2013 at 1:00 AM. The log was downloaded on September 16, 2018 at 8:00 PM.
99
10EventLogDataAnalysis
Objectives

Obtain the times-to-failure and times-to-repair distributions of each subsystem. Use overlay plots to compare the failure behaviors and the repair behaviors of the subsystems.
Solution
To obtain the failure distribution and repair distribution of each subsystem, you will use the event log folio to analyze the information downloaded from the data logger. To create a new event log folio, you choose Insert > Folios > Event Log.
Once the folio is created (Auto Data Event Log in the sample project), the next step is to configure the data sheet. First, you click the Other page icon on the control panel, as shown next.
In the Time to Failure/Repair In area, you select Hours so that when the software converts the maintenance logs to failure and repair data, the unit of time will be hours. Because all of the subsystems will stop accumulating age if the equipment fails for any reason, you clear the Use OTSF column check box. If this check box were selected, the OTSF (operates through system failures) column would be displayed in the data sheet and you would have the option to specify which
100
10.1EventLogData
subsystem(s) will continue to accumulate age when the system is down due to the failure of another subsystem. You are analyzing only one piece of equipment, so you also clear the Use System column check box. If this check box were selected, the System column would be displayed in the data sheet and you would have the option to enter events for multiple systems. Next, you click the Analysis page icon, as shown next.
For both the failure and repair distributions, you select the Prefer RRX if sufficient data option and the 2P-Weibull distribution. These will be the settings that the software will use to automatically analyze the failure and repair data that are extracted from the event log.
101
Next, you enter the information from the data logger into the data sheet, as shown next.
Event Log Data Entry: In the event log folio, the F=Failure/E=Event column indicates whether theoccurrence was afailure event (F) orageneral event (E).Ageneralevent represents an activity that brings the system down butisnotdirectly relevant to thereliabilityof theequipment,suchas preventivemaintenance,routineinspectionsandthelike.
The Level 1, 2, 3 and 4 columns allow you to indicate the subsystem or component that was responsible for the event. The level numbers represent the hierarchy of the subsystems, where Level 1 is the highest level in the system configuration at which you might want to perform the analysis.
102
10.1EventLogData
Once data entry is complete, you click the Main page icon on the control panel, as shown next.
In the Levels to Analyze area, you select the Level 1 check box because only one level was recorded in the data set. In the Analyze Failures and Events area, you select the Separately option. This setting tells the software to fit a separate distribution to the E events (if any), instead of combining both types of events together in the same analysis. Because this data set contains only failure events (F), this setting will not affect the analysis results, but it will provide more appropriate labeling for the extracted life data (i.e., will show [F] for failures instead of [C] for combined events). In the System area, you enter the dates/times when the observation period started and ended. For the start date/time, you enter January 2, 2013 at 1:00 AM. For the end date/time, you enter September 16, 2018 at 8:00 PM.
103
Because the start date/time is also the exact calendar date that the equipment started operating, you select the System is new on start date check box.
SystemisNewonStartDate:Thedates/timesthatyouprovideintheSystemareaofthecontrol panelareusedtodeterminewhetherthefirsteventtimeisrecordedascompleteorcensoreddata. If the system started operating on the same calendar date as the start date of the observation period (i.e., If the system was new when you started recording data), then the data point is complete (F) because the exact timetoevent is known. If the system was not new, then the data point is censored (S) because you don't know the amount of time that the system operated before theobservationperiodbegan.
To specify the shift schedule of the equipment, you click the Shift Pattern icon on the control panel.
The equipment operates for 24 hours a day, 7 days a week, so you set up the shift schedule as shown next.
Shift Schedules: In the case of equipment that does not run on a 24/7 basis, the Shift Pattern feature allows the analysis to take into consideration the periods when your system is not in use. Thisensuresaccuratecalculationsforthefailureandrepairtimes.
Once the settings are finalized, you analyze the data (i.e., convert the logs into failure and repair data, and fit distributions to the data sets) by clicking the Calculate icon on the control panel.
104
10.1EventLogData
After the folio is calculated, you click the Show Analysis Summary (...) button on the Main page of the control panel, as shown next.
The following report shows the values of the parameters of the failure and repair distributions for each subsystem, as well as the calculated uptime and downtime of each subsystem.
To compare the probability plots for all subsystems, you transfer the data from the event log folio to a Weibull++ standard folio by choosing Event Log > Transfer Life Data > Transfer Life Data to New Folio.
This creates a new folio that contains separate data sheets for the failure and repair data of each subsystem (Event Log - Extracted Data in the sample project).
105
You then calculate each data sheet using the 2P-Weibull distribution and the RRX parameter estimation method. The following data sheet shows the analysis of the failure data for subsystem 1.
Tip:Inthedatasheetnames,the[F]tagindicatesananalysisofthefailures,while[E]isananalysisofthegeneral events(notethatacombinedanalysiswoulddisplay[C]).The[FD]tagindicatesthatthedatasetisforthefailure distribution,while[RD]isthedatasetfortherepairdistribution.
106
10.1EventLogData
Once all the data sets are calculated, you create an overlay plot by choosing Insert > Report and Plots > Overlay Plot.
When prompted to select which data sets to plot, you select all the data sheets that contain the failure data of the subsystems (i.e., all the data sheets that have the [FD] tag). The following overlay plot (Event Log - Failures in the sample project) shows the failure distributions plotted on the same Weibull probability plot.
You then create a second overlay plot that displays all the repair distributions on the same plot (i.e., all the data sheets that have the [RD] tag). The following overlay plot (Event Log - Repairs in the sample
107
project) shows the repair distributions plotted on the same Weibull probability plot.
108
ALTAwithOneStress
ReliaSofts ALTA 8 software provides a comprehensive toolset for accelerated life test planning and quantitative accelerated life testing data analysis, plots and reporting. In this chapter, you will use the ALTA standard folio to perform the simplest kind of accelerated life testing data analysis. One accelerated stress will be applied during the test, and you will extrapolate the product's failure behavior at normal conditions from the life data obtained at the accelerated levels.
11
ALTAwithone stresstype ALTAQuick CalculationPad (QCP)
After the failure times are fitted to an ALTA model (i.e., a life-stress Overlayplots relationship combined with a lifetime distribution), you will use the model to calculate a variety of reliability metrics (i.e., mean time to failure, B2 life and acceleration factor) and create an overlay plot to compare a product that was tested at the use stress level to a product that was tested at accelerated levels. Watchavideoofthisexampleat http://www.ReliaSoft.tv/alta/startguide/alta_qsg_11.html.
11.1AcceleratedLifeTestingDataAnalysisOneStressType
While your company was evaluating bulb A (see Chapter 2), the purchasing group found a new supplier that provides a less expensive bulb C. The new supplier claims that the bulb is more reliable, and to support this claim, they have provided data from an accelerated life test using increased temperature. According to the supplier, the tests were done at 40 and 55 degrees Celsius (i.e., 313 and 328 kelvins) and resulted in the failure/suspension times shown in the following table.
109
11ALTAwithOneStress
State(F=Failedand S=Suspended) F F F F S S S F F F F F F
Time(Hr) 105 170 224 278 350 350 350 38 62 82 103 128 165
Temperature(K) 313 313 313 313 313 313 313 328 328 328 328 328 328
You are asked to use this data set to compare bulb C to bulb A, assuming a normal operating temperature of 25 degrees Celsius (i.e., 298 K).
Objectives

Calculate the mean time to failure and B2 life of the new suppliers bulb. Estimate the acceleration factor at 313 K. Generate an overlay plot that compares the unreliability vs. time for bulb A and bulb C.
Solution
You start by choosing Insert > Folios > ALTA Standard Folio to create a new folio.
110
When prompted to specify the data type, you select My data set contains suspensions. You also use the Units drop-down list to indicate that the time values in the data sheet will be entered in hours, as shown next.
You click Next. On the second page of the setup window, you select Temperature as the stress type and enter 298 for the use level temperature. The setup is shown next.
111
11ALTAwithOneStress
After the folio is created (Bulb C - Supplier Data in the sample project), you enter the failure/ suspension times and the stress levels at which those times were obtained. For example, rows 5 through 7 in the data sheet shown next are units that were still operating after 350 hours at a test temperature of 313 K, while row 13 shows the last failure recorded at 165 hours for the units tested at 328 K.
Tip: Temperature values must always be entered in absolute units (e.g., kelvins or degrees Rankine). To convert temperaturevaluesinthedatasheetthatwereenteredinCelsiusorFahrenheittoabsoluteunits,chooseALTA> Options>ConvertStressValues. An ALTA model is a combination of a life-stress relationship and life distribution. From the Model dropdown list, you select to use the Inverse Power Law relationship because the failure mode under consideration is filament burnout. In the submenu that appears, you select IPL-Weibull to use the Weibull distribution with the selected relationship. To fit the selected model to the data, you click the Calculate icon on the control panel.
112
Once the model is fitted, the Analysis Summary area of the control panel displays its parameters and other relevant information, as shown next.
To create a probability plot to visualize the analysis results, you click the Plot icon on the control panel.
On the control panel of the plot sheet, you choose the Probability - Weibull plot, as shown next (with annotations added via RS Draw to make the plot easier to interpret).
The line to the left in this plot shows the unreliability calculated for the data obtained at 328 K. The line in the center shows the unreliability at 313 K, and the line to the right shows the extrapolated unreliability at the specified use stress level (i.e., 298 K).
113
11ALTAwithOneStress
To calculate the specific results of interest, you open the Quick Calculation Pad (QCP) by clicking the icon on the control panel.
You select to calculate the Mean Life and use the Units drop-down list to make sure the results will be returned in hours. You also make sure that the correct stress is defined in the Stress field. The results appear as shown next.
Bulb Cs mean life is approximately 1,122 hours, which is higher than the 906-hour mean life you calculated for bulb A. See page 10 in Chapter 2. You use the QCP to solve the remaining metrics:
To estimate the B2 life of bulb C, you select to calculate BX% Life and enter 2 for the BX% Life At input. The result is approximately 229 hours, which is lower than the 377-hour B2 life of bulb A. See page 10 in Chapter 2. To estimate the ratio between the life of the bulb at an accelerated stress level to its life at the normal stress level, you select to calculate the Acceleration Factor. In the Stress field, you enter 298, while in the Accelerated Stress field, you enter 313. The result is an acceleration factor of 3.539260 or about 3.5. This means that bulb Cs life is about 3.5 times longer at the use stress level than it is at a higher stress level of 313 K.
114
Finally, to compare the unreliability vs. time for bulb A and bulb C, you create an overlay plot by choosing Insert > Reports and Plots > Overlay Plot.
In the Select Data Sheets window, you select the Weibull++ data sheet with the combined data for bulb A and the ALTA data sheet with the supplier data for bulb C (Bulb A - Combined Data and Bulb C Supplier Data in the sample project). In the overlay plot, (Bulb A vs. Bulb C in the example project), you switch the plot type to a Use Level Probability Weibull, as shown next (with annotations added via RS Draw to make the plot easier to interpret).
As you can see, the slopes of the two lines are different, and they intersect near the middle of the plot. This shows that at earlier times bulb A has a higher reliability than bulb C, but at later times bulb Cs reliability is higher.
115
ALTAwithTwoStresses
In many practical applications, the life of a product is a function of two or more variables. To support this type of analysis, ALTA includes two-stress models and multivariable models that allow you to analyze multiple accelerating stresses. In this chapter, you will perform a test with two accelerated stresses and use the ALTA standard folio to analyze the resulting data.
12
ALTAwithtwo stresstypes ALTAQCPand plots
Watchavideoofthisexampleat http://www.ReliaSoft.tv/alta/startguide/alta_qsg_12.html.
12.1AcceleratedLifeTestingDataAnalysisTwoStresses
After analyzing the accelerated life test data provided by the supplier of the new bulb, you begin to investigate further and discover that the primary cause of the bulb failures, which is filament burnout, may be influenced by both temperature and voltage. Therefore, you request permission to perform an in-house accelerated life test for the bulbs, which considers both stress factors. You are allocated 40 bulbs for testing purposes and have permission to use the lab facilities for a maximum of 400 hours (any failures after 400 hours will be suspended). After studying the design limits of the bulb, you create the following plan:

11 bulbs will be tested at 318 K and 120 V 10 bulbs will be tested at 318 K and 125 V 9 bulb will be tested at 333 K and 120 V 10 bulbs will be tested at 333 K and 125 V
117
12ALTAwithTwoStresses
You obtain the following data from the test: Number ofUnits 1 1 1 1 1 1 1 4 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 State (FFailure/SSuspension) F F F F F F F S F F F F F F F F F F F F F F F F F F F F F F F F F F F F F StateEnd Time(Hr) 205 256 259 296 326 352 378 400 103 130 148 163 176 189 202 216 232 77 97 111 122 132 142 151 162 174 192 33 42 48 53 57 61 66 70 75 83 Temperature (K) Voltage (V)
318
120
333
318
125
333
118
Objectives

Calculate the mean time to failure and B2 life of the bulb. Estimate the acceleration factor at 318 K and 125 V. Generate an overlay plot that compares the use level probability plots of three data sets: combined data on bulb A, supplier data on bulb C and in-house test data on bulb C.
Solution
You start by choosing Insert > Folios > ALTA Standard Folio to create a new folio.
When prompted to specify the data type, you select the following check boxes:

My data set contains suspensions (right-censored data) I want to enter data in groups
You also use the Units drop-down list to indicate that the time values in the data sheet will be entered in hours, as shown next.
119
You click Next. On the second page of the setup window, you select Temperature and Voltage as the stress types. You enter 298 for the use level temperature and 120 for the use level voltage. The setup is shown next.
Once the folio is created (Bulb C - In-House Data in the sample project), you enter the data from the test into the folio data sheet, as shown next.
120
The next step is to select the analysis settings. On the control panel, you select the General Log-Linear model. In the submenu that appears, you select GLL-Weibull to use the Weibull distribution with the selected model. To specify which stress types will be considered in the analysis, you click the Select Stress Columns link. In the Select Stress Columns window, you select the check boxes for temperature and voltage. To specify the transformations for the GLL model, you click the Stress Transformation link. In the Stress Transformation window, you choose the logarithmic transformation for both temperature and voltage (because the failure mode under consideration is filament burnout), as shown next.
Stress Transformation: The general loglinear (GLL) model is a multivariable relationship model
thatsupportscaseswheremorethantwoacceleratedstressesneedtobesimultaneouslyanalyzed. In ALTA, the GLL model can analyze life as a function of up to 8 stress types and, if desired, the relationship may be modified through the use of stress transformations. You have the option to choose one transformation for each stress type that is selected to be used in the analysis. This allows you to modify the GLL model in terms of a lifestress relationship that is appropriate for the stresstype. To fit the selected model to the data, you click the Calculate icon on the control panel.
121
Once the model is fitted, the Analysis Summary area of the control panel displays its parameters and other relevant information, as shown next.
To create a probability plot to visualize the analysis results, you click the Plot icon on the control panel.
On the control panel of the plot sheet, you choose the Probability-Weibull plot. The resulting plot shows the lines for the accelerated stress levels and the line for the use stress level. The lines are parallel, indicating a common beta, as shown next (with annotations added via RS Draw to make the plot easier to interpret).
122
To solve for the requested reliability metrics, you open the Quick Calculation Pad by clicking the QCP icon on the control panel.
You select to calculate the Mean Life and use the Units drop-down list to make sure the results will be returned in hours. The mean life of the bulb is estimated to be 1,131 hours, as shown next.
You use the QCP to solve the remaining metrics:
To estimate the B2 life, you select to calculate the BX% Life and enter 2 for the BX% Life At input. The result is about 526 hours. To estimate the ratio between the life of the bulb at accelerated stress levels to its life at the normal stress level, you select to calculate the Acceleration Factor. In the Accelerated Stress field, you enter 318 for the temperature and 125 for the voltage, as shown next. The result is a ratio of about 8.494413.
123
Next, you compare the current analysis to two other analyses. You create an overlay plot by choosing Insert > Reports and Plots > Overlay Plot.
When prompted to select which data sets to plot, you choose the following:
The ALTA folio data sheet containing the data set for bulb C from the in-house test. (Bulb C - InHouse Data in the sample project.) The ALTA folio data sheet containing the data set for bulb C from the supplier. (Bulb C - Supplier Data in the sample project.) The Weibull++ folio data sheet containing the combined data set for bulb A. (Bulb A - Combined Data in the sample project.)
The following overlay plot (Bulb A vs. Bulb C (In-House and Supplier) in the sample project) shows the analyses plotted on the same Use Level Probability Weibull plot (with annotations added via RS Draw to make the plot easier to interpret).
As you can see, the lines for bulb C based on the in-house test and bulb A are parallel, while the line for bulb C based on the suppliers data has a different slope.
124
To show the confidence bounds, you right-click the plot sheet and select Confidence Bounds. You then select Two-Sided bounds and enter 90 for the confidence level. The result shows that the bounds on the suppliers analysis are much wider compared to the bounds on the in-house analysis.
This indicates that the results of the in-house analysis may provide more precise estimates, allowing you to perform a better assessment of bulb Cs reliability against bulb A.
125
ALTAwithTime DependentStressProfiles
With ALTA PRO, you can analyze data obtained from products whose stress levels vary with time. The software allows you to define stress profiles that describe how stress levels increase/decrease with time. When you select to use the Cumulative Damage (CD) life-stress relationship, which takes into account the cumulative effect of the applied stresses, the profiles you've defined become available for use in your analysis. In this chapter, you will perform accelerated life testing data analysis on a product with timedependent use stress conditions.
13
ALTAwithtime dependentstress types Stressprofiles
13.1TimeDependentStress
Thus far, your analysis of the accelerated life test data for the new bulb C has assumed a constant stress level under normal use conditions. (See Chapter 11 and Chapter 12.) You later learn from the design team that the actual usage conditions are more likely to vary with time. You record the actual operating conditions of the bulb for a period of 72 hours. Specifically, you notice the following patterns:

The temperature alternates between 290 K and 305 K every 12 hours. The voltage switches between 119 V and 123 V in the following pattern: OperatingHours 0to10thhour 10thto12thhour 12thto22ndhour 22ndto25thhour At25thhouruntilendofobservationperiod Voltage 119V 123V 119V 123V 123V
127
13ALTAwithTimeDependentStressProfiles
Objectives
Calculate the mean time to failure and B2 life of the bulb, when the analysis considers the timedependent use stress conditions. Use the B2 life of the bulb to estimate the acceleration factor at 318 K and 125 V. Generate an overlay plot that compares the new analysis (time-dependent use stress levels) versus the previous analysis (constant stress levels).
Solution
The first step is to create stress profiles in ALTA that reflect the time-dependent operating conditions that the bulb will experience.
StressProfiles:Astressprofiledescribeshowstresslevelsvarywithtime.Itconsistsofaseriesof time segments, each with a specified constant or timedependent stress level. Stress profiles are used with the cumulative damage lifestress model to perform an analysis that takes into account timevaryingstresslevelsundertestingand/orduringnormaluse.
You create a stress profile by choosing Insert > Tools > ALTA Stress Profile.
Once the profile is created, you change its default name to Use Temperature by right-clicking the new profile in the current project explorer and choosing Rename on the shortcut menu. This will make the profile easier to recognize when you need to select it within the analysis. In the data sheet of the stress profile folio (Use Temperature in the sample project), the start time for the first segment is always set to zero. The start times for subsequent segments are calculated automatically by the software. Thus, you will enter only the segment end times and corresponding temperature values. After entering the data, you select the Repeat cycle option on the control panel. This tells the software to treat the entire pattern of segments as a repeating cycle.
128
The following picture shows the completed setup.
To save your changes and validate the profile, you click the Validate Stress Profile icon on the control panel.
To generate a plot that shows the pattern of the segments, you click the Plot icon on the control panel.
You can view a The following plot shows the stress profile for temperature (with the scaling adjusted to Y = 0 to 400 and X = 0 to 100). It illustrates how the normal usage temperature varies with time.
129
Next, you create a second stress profile for voltage and rename it to Use Voltage. In the data sheet of the new stress profile (Use Voltage in the sample project), you enter the segment end times and corresponding voltage values. After entering the data, you select the Continue from last stress option on the control panel. This tells the software that all the times after the last segment will use the stress value defined in the last segment. In other words, once the projector reaches 123 V for the second time, it stays at that value for as long as the projector continues to operate. The completed setup is shown next.
As before, you click the Validate Stress Profile icon on the control panel to validate the data sheet entries and save your changes.
130
You also generate a plot to show how the normal usage voltage varies with time, as shown next (with the scaling adjusted to Y = 0 to 150 and X = 0 to 30).
Next, you duplicate the folio that contains the in-house data for Bulb C (Bulb C - In House Data in the sample project). You do this by right-clicking the folio in the current project explorer and choosing Duplicate on the shortcut menu. Once the duplicate folio is created, you rename it to distinguish it from the original folio (Bulb C - In House Data - Stress Profiles in the sample project). For this analysis, you change the ALTA model to the cumulative damage model in order to use the timedependent stress profiles to analyze the data. On the control panel, you choose Cumulative Damage > CD-Weibull. You then click the Select Stress Columns link and make sure that the check boxes for temperature and voltage are still selected. As with the general log-linear model in ALTA, the cumulative damage model allows you to define the transformation that will be used for each stress type. You click the Stress Transformation link and make sure that the transformations are still set to logarithmic for both stress types (because the failure mode under consideration is filament burnout).
131
To change the constant values that were defined in the previous analysis to the time-dependent stress profiles, you click the Set Use Stress link. In the Use Stress Level window, you select the Profile check box next to each stress and choose the corresponding stress profile from each drop-down list, as shown next.
You click the Calculate icon on the control panel to analyze the data.
To solve for the requested reliability metrics, you open the Quick Calculation Pad by clicking the QCP icon on the control panel.
You select to calculate the Mean Life and use the Units drop-down list to make sure the results will be returned in hours. In the Stress field, you make sure that the use stress is obtained from the stress profiles.
132
The mean life of the bulb is then estimated to be 424 hours, as shown next.
You continue using the QCP to solve the remaining metrics:
To estimate the B2 life, you select to calculate the BX% Life. In the Stress field, you select the Profile check box next to each stress and select the corresponding stress profile from each dropdown list. Next, you enter 2 for the BX% Life At input. The result is about 205 hours. You select to calculate the Acceleration Factor.

In the Stress field, you make sure that the Profile check box next to each stress is selected. In the Accelerated Stress field, you clear the Profile check boxes because the accelerated stress values are constant. You then enter 318 for the temperature and 125 for the voltage. In the Mission End Time field, you enter 205.
The result is a ratio of 7.328050. This means that bulb Cs life is about 7.33 times longer at the use stress level than it is at higher stress levels of 318 K and 125 V.
133
Next, you compare the new analysis with the previous analysis. You create an overlay plot by choosing Insert > Reports and Plots > Overlay Plot.
When prompted to select which data sets to plot, you choose the data sheet with the time-dependent stresses and the data sheet with the constant stresses (Bulb C - In House Data - Stress Profiles and Bulb C - In House Data in the sample project). On the control panel of the plot sheet, you select the Unreliability vs. Time plot. The following overlay plot (Time-Dependent Stress vs. Constant Stress in the sample project) shows the result of the assumption of a constant stress against time-dependent stress (with annotations added via RS Draw to make the plot easier to interpret).
If the usage conditions of the bulb vary with time, then the assumption of a constant stress over-estimates the life of the bulb. In this case, the analysis with the time-dependent stress profiles provides a more accurate model of the operating conditions of the bulb and, therefore, more precise estimates.
134
AcceleratedDegradation Analysis
In Chapter 3, we illustrated how degradation analysis can reduce the time required to obtain a products life data. To reduce testing time even further, the degradation can be measured at accelerated stress levels. This type of approach is known as accelerated degradation analysis. In this chapter, you will work with degradation measurements obtained at accelerated stress levels and use the ALTA degradation folio to estimate the products life characteristics under normal stress conditions.
14
ALTADegradation folios Overlayplots
14.1AcceleratedDegradationAnalysis
The LED lamp designers that you worked with before (Chapter 3) want to perform another test to capture data about the degradation in the light output over time. But this time, instead of 2 months, they can reserve the lab for only 2 weeks (336 hours) so they decide to use an accelerated test. For reference, they give you the following information:
The normal stress level of the lamp is 120 volts, but higher voltage levels are known to accelerate the progression of the same failure modes that occur under normal use conditions. The degradation of the lamp is measured in terms of how the light output decreases over time. The lamp is considered failed when its light output reaches 50% of its design output of 1,000 lumens.
The plan is to test 10 lamps at 128 volts and another 10 at 138 volts. The highest voltage level is expected to accelerate the degradation in the light output without introducing foolish failures (i.e., failures that would never occur under normal use conditions). The light intensity of the lamps will be recorded at 100, 150, 200 and 250 hours.
135
14AcceleratedDegradationAnalysis
You obtain the following measurements from the test. LuminosityMeasurementsat128Volts Inspection Time(Hr)
100 150 200 250
Lamp 01
979 908 867 826
Lamp 02
969 918 877 836
Lamp 03
1010 908 847 796
Lamp 04
857 745 683 627
Lamp 05
898 775 724 683
Lamp 06
816 785 745 714
Lamp 07
949 908 857 836
Lamp 08
887 826 775 745
Lamp 09
887 816 745 704
Lamp 10
928 867 816 785
LuminosityMeasurementsat138Volts Inspection Time(Hr)

100 150 200 250
Lamp 11
768 712 680 648
Lamp 12
760 720 688 656
Lamp 13
792 712 664 624
Lamp 14
672 584 536 492
Lamp 15
704 608 568 536
Lamp 16
640 616 584 560
Lamp 17
744 712 672 656
Lamp 18
696 648 608 584
Lamp 19
696 640 584 552
Lamp 20
728 680 640 616
Objectives

Use a Degradation vs. Time plot to see how the luminosity of the lamps degrades over time. Generate an overlay plot that compares the probability plots for the accelerated degradation analysis and the standard degradation analysis in Chapter 3.
Solution
You create an ALTA degradation folio by choosing Insert > Folios > ALTA Degradation.
136
When prompted to define the stress, you select Voltage and then enter 120 for the use stress level.
Once the folio is created (Accelerated Degradation - LED Lamp in the sample project), you enter the degradation measurements into the folios data sheet. The next step is to specify how the failure times will be extrapolated from the luminosity measurements. Based on your engineering knowledge, you select Power from the drop-down list in the Degradation Model area. For the lamps Critical Degradation (i.e., the luminosity level at which a lamp is considered failed), you enter 500 lumens.
137
Next, you need to specify how the extrapolated failure times will be analyzed. In the ALTA Model area, you select IPL-Weibull. The following picture shows the completed setup.
To analyze the data set, you click the Calculate icon.
The software will first extrapolate the times at which the degradation measurements will reach the point of critical degradation at the accelerated levels, and then it will perform a separate accelerated life testing data analysis on the extrapolated data. To view a plot that shows how the luminosity for each unit in the sample degrades over time, you click the Plot icon on the control panel and select the Degradation vs. Time (Linear) plot. The resulting plot is
138
shown next (with the scaling adjusted to Y = 450 to 1,000 lumens). The horizontal line near the bottom of the plot marks the luminosity level at which the lamp is considered failed.
You know that in accelerated life testing data analysis, the farther the accelerated stress level is from the normal stress level, the greater the uncertainty in the extrapolation. To compare the results obtained from the accelerated degradation test to the results obtained from the standard degradation test, you generate an overlay plot by choosing Insert > Reports and Plots > Overlay Plot.
When prompted to select which data sets to plot, you select the data sheet containing the ALTA degradation analysis and the data sheet containing the Weibull++ degradation analysis (Accelerated Degradation - LED Lamp and Degradation - LED Lamp in the sample project).
139
On the control panel of the plot sheet, you switch the plot type to a Use Level Probability Weibull plot. The following overlay plot (Degradation Analysis Comparison in the sample project) shows the results of the two analyses on the same plot (with 90% two-sided confidence bounds on reliability).
The overlay plot shows that the probability lines have similar slopes; however, the bounds on the accelerated degradation analysis are much wider compared to the bounds on the standard degradation analysis. Nevertheless, you were able to obtain reasonable results within the allotted test time.
140
RELIASOFT CORPORATION END-USER LICENSE AGREEMENT

This ReliaSoft Corporation End-User License Agreement is a legal agreement between You and ReliaSoft Corporation for ReliaSoft Corporation's software products, which includes computer software and associated media and printed materials as well as on-line or electronic documentation and any software updates or upgrades, henceforth collectively called the Application. By installing, copying or otherwise using the Application, You agree to be bound by the terms of this End-User License Agreement and represent and warrant that You are authorized to do so. If You do not agree to the terms of this End-User License Agreement, do not install or use the Application and promptly return the Application to the vendor from which you obtained it within fourteen (14) calendar days from the date of purchase for a full refund. This Agreement is effective from the earlier of the date You open the sealed software package, if applicable, or install the Application, or fourteen (14) calendar days from the date the Application was purchased or otherwise obtained from ReliaSoft Corporation, and continues until expressly terminated by ReliaSoft Corporation. This Application is protected by copyright laws and international copyright treaties, as well as other intellectual property laws and treaties. This Application is licensed, not sold.
DEFINITIONS
1.1
1.2
1.3 1.4 1.5
1.6
Access. Means connected to or using the Application in any manner, either directly or indirectly through any middle tier application or system. Activate. Means the act of enabling the Application to operate on a specific computer by registering the Application license with the ReliaSoft License Server and activating the Application for operation. Location. Means the postal address of the site where the Application shall be installed. Location shall mean a single physical location. Named User(s). Means the individual(s) permitted to Access the Application, and it refers to specific individuals. ReliaSoft License Server. Means the ReliaSoft server(s) that provides license activation, registered end-user management, license management and license compliance enforcement services for the Application. You. Means the licensee or the individual acquiring the license of this Application and, in either case, the affiliates, employees, contractors and agents of said licensee or individual. Grant. Subject to the license types defined in Section 3 hereof, You are granted a personal, non-exclusive and limited license to install, Activate and Access the Application under the license type and in the quantities for which You have obtained a license. You may install and Activate the Application onto designated equipment at the specified Location(s), solely as necessary to use the Application in accordance with this Agreement. You may not
LICENSE GRANT
2.1
2.2
2.3
distribute, sublicense, copy (except as expressly permitted in this Agreement), transfer (except as expressly permitted in this Agreement), nor give the Application to any third party. Ownership. The Application and the information it contains, including any updates, upgrades, modifications and derivative works, and all copies of them, are the property of ReliaSoft Corporation (or its licensors), and title remains with ReliaSoft Corporation (or ReliaSoft Corporation's licensors). All applicable rights in patents, copyrights and trade secrets in the Application are and will remain with ReliaSoft Corporation (or ReliaSoft Corporations licensors). No title to, or ownership of, the Application or the information contained therein is transferred to You. ReliaSoft Corporation reserves all rights not explicitly granted in this Agreement. Rights to Developments. This Agreement also governs your use of any enhancements, data, or information provided by ReliaSoft Corporation while performing any technical, training or consulting services. Any ideas, know-how, techniques and software developed by ReliaSoft Corporation, including any enhancements or modifications made to the Application by any person, are the property of ReliaSoft Corporation.
INSTALLATION AND USE

ReliaSoft Corporation licenses any Application under this Agreement on a per user basis. In other words, ReliaSoft Corporation licenses an Application to each user and not to a particular computer. This Agreement grants You the following rights, depending on the license You have obtained from ReliaSoft Corporation: 3.1 Single User License. The Single User License is licensed to one individual Named User. This is a perpetual license. Each software package includes the installation media and one copy of the Application documentation per license, either in physical/hard copy format and/or electronic format. The Named User may install, Activate and Access the Application on up to three computers (i.e., multiple Locations), such as a work computer, a home computer and a portable computer, provided that the Named User is the only person who uses this Application, at any time, in any of these Locations. In the event that installation and activation of the Application is performed on three computers, the Named User will be required to deactivate at least one installation before being allowed to Activate the Application again. There is a limit of 6 activations allowed per calendar year. The Single User License cannot be installed on multi-user capable environments such as Windows, Terminal or Citrix servers. Use of the Application on any of the computers by any user other than the Named User who has licensed the Application is strictly prohibited. Such other users must obtain a separate license of the Application from ReliaSoft before any use. 3.2 Standard Network License. The Standard Network License is licensed to Named Users in an organization on a single network of computers at a specified Location. This is a perpetual license. The software package includes the installation media and one copy of the Application documentation per license, either in physical/hard copy format and/or electronic format. Note: This is not a Concurrent Network License.
3.3
This license provides for a limited number of Named Users, dependent on the quantity of seats licensed, to install, Activate and Access the Application, with a single activation allowed for each Named User. For example, if You purchase a ten (10) seat Standard Network License, only ten (10) specific Named Users are allowed Access. If an eleventh (11th) user wishes to have Access to the Application, an additional seat must be licensed. In the event that installation and activation of the Application is required on a new computer for a Named User, the Named User will be required to deactivate the active installation before being allowed to install and Activate the Application on the new computer. The Standard Network License allows for installation on multi-user capable environments such as Windows, Terminal or Citrix servers. Under this license, no other copies can be installed on non-network computers, such as portable computers, home computers, etc. Concurrent Network License. The Concurrent Network License is licensed to an organization with multiple regular and occasional Named Users on a single network of computers at a specified Location. This is a perpetual license. The software package includes the installation media and one copy of the Application documentation per license, either in physical/hard copy format and/or electronic format. Concurrent Network Licenses are specified by two attributes: (i) Number of Named User Activations, which is the maximum number of computers on which a Named User can Activate the Application (i.e., make it available for operation). More than one activation is allowed for each Named User (e.g., to install the Application on both a work computer and a portable computer), but each activation is counted toward the maximum available under the license. (ii) Number of Concurrent Access Licenses (or CALs), which is the maximum number of computers that can Access the Application at any given time. For example, for a Concurrent Network License with thirty (30) Named User Activations and ten (10) CALs, the Application can be installed and activated on up to 30 computers (e.g., by 30 Named Users each activating the Application on only one computer, by 15 Named Users each activating the Application on two computers, etc.), and only ten (10) of those computers will be able to Access the Application at the same time. If a user attempts to Access the Application from an eleventh (11th) computer, the user must wait until one of the other ten users ends their session and releases the CAL. The Concurrent Network License allows for installation on multi-user capable environments such as Windows, Terminal or Citrix servers. Under this license, Named Users are allowed to check out a CAL from the ReliaSoft License Server for a specified period of time, effectively being guaranteed Access to the Application while simultaneously reducing the number of available CALs for the rest of the user pool until the CAL is checked back into the ReliaSoft License Server or otherwise released. This ability to check out a CAL from the ReliaSoft License Server is only available for Applications designated as Major Version 8 and above (e.g., 8.0.1, 8.0.2, etc). This type of license requires that you provide/allow internet access from the Application to the ReliaSoft License Server at https://Validate.ReliaSoft.org. Concurrency authorization
3.4
3.5
3.6
and control is managed and provided by this server. You may need to modify and adjust your internet and/or firewall settings and/or set the aforementioned server as a trusted location in order to use the Application. This requirement for connectivity to the ReliaSoft License Server for concurrency authorization and control only applies for Applications designated as Major Version 8 and above (e.g., 8.0.1, 8.0.2, etc). Asset-based License (only applies to Orion eAPI software). The Asset-based License is licensed to all direct employees of a corporate entity within one or more specified Sites, depending on the license obtained. This license excludes subcontractors, suppliers or temporary employees, unless they are using the Application inside an eligible Site on a computer owned by that Site. This is a perpetual license. Under this license, an unlimited number of Named Users from the licensing Site or Sites is allowed to Access the Application; however, the license restricts the number of physical assets (commonly referred to as Asset TAGs) that can be configured and managed within the Application. For this license, a Site is defined as a distinct physical address/site of a corporate entity, where the physical assets being managed reside (e.g., a refinery or a mine or a power station). The license specifically excludes broader geographical areas, such as cities or states/provinces (where multiple Sites might exist). Unlimited User License. The Unlimited User License is licensed to all direct employees of a corporate entity within a specified Site, Division, or Corporation, depending on the type of license obtained. This license excludes subcontractors, suppliers or temporary employees, unless they are using the Application inside an eligible facility on a computer owned by that facility. This is a perpetual license. Site-Wide License: Covers all direct employees working at a distinct physical address/site of a corporate entity. A Site-Wide License specifically excludes broader geographical areas, such as cities or states/provinces (where multiple sites might exist). Division-Wide License: Covers all direct employees affiliated with a unique business unit of a corporate entity organized in multiple divisions/business units. Examples of divisions are the Aerospace Division of a large conglomerate or the Brakes Division of an automotive corporation. A Division-Wide license might additionally be restricted to a specific geographical area to facilitate licensee requirements (for example, the Aerospace Division of a conglomerate in North America). Corporate-Wide License: Covers all direct employees of a corporate entity. This license entitles the specified corporate entity at the specified Site/Division/Corporation to make and distribute copies of the Application in executable code in its entirety, as supplied by ReliaSoft Corporation, OR install and distribute the Application over a network. The software package includes the installation media and one copy of the Application documentation per license, either in physical/hard copy format and/or electronic format. The Unlimited User License allows for installation on multi-user capable environments such as Windows, Terminal or Citrix servers. Rental License. Rental licenses are available for the Single User, Standard Network and Concurrent Network licenses. The license terms are the same as the corresponding license
type being rented except that the license expires according to the term purchased. The rental fee will not be prorated or refunded if license rental is cancelled or not used in full. A standard non-expiring license may be purchased at any time at the current Application list price. 3.7 Evaluation, Demo, No Save Demo, Beta, or Otherwise Expiring or Non-Expiring Licenses of any Type Provided for Evaluation Purposes. These are Evaluation/Not for Sale licenses and, notwithstanding any other section of this Agreement, You may not use the Application for commercial purposes, nor sell or otherwise transfer it for value. Commercial purposes include the use of the Application in creation of publicly distributed computer software. You are acquiring only the limited right to use a single copy of the Application solely for evaluation purposes and for a limited time. ReliaSoft Corporation can terminate this Agreement at any time. You have no rights under any sections of this Agreement other than those specifically allowed for within this section (Section 3.7). THE APPLICATION IS LICENSED AS IS with no representations or warranties of any kind. Note: As the section title implies, users of any license provided by ReliaSoft Corporation for evaluation purposes (even if it is, for example, a Single User or Network License), shall abide by the rules set forth in this section. 3.8 Complimentary License. Complimentary Licenses have the same installation and use rights as their corresponding purchased licenses (e.g., a Complimentary Single User License provides the same installation and use rights as the Single User License). However, You have no rights under any other sections of this Agreement other than those specifically allowed for within this section (Section 3.8). THE APPLICATION IS LICENSED AS IS with no representations or warranties of any kind and license transfers are NOT allowed. Regardless of the license type You have obtained from ReliaSoft Corporation, You may access shared data storage files or databases relating to the Application across a wide area network (WAN) or a virtual private network (VPN), instead of a local area network (LAN), as long as such access is permitted by your policies and authorized by You via appropriate Application and network configuration settings. Although such access is not prohibited by ReliaSoft Corporation, you acknowledge that WAN and VPN data transfer rates vary widely and could lead to decreased and unpredictable performance (as compared to a LAN) and therefore disclaim any expectation of Application performance should you choose to a) configure any part of the Application, b) use the Application, or c) access shared data storage files in a distributed environment over a WAN or VPN. Such disclaimer does not apply to any Application designated and licensed as a native Web-based Application by ReliaSoft Corporation.
SOFTWARE TRANSFERS
4.1
Single User License. You may permanently transfer all of your rights related to a Single User License under this Agreement, provided that You retain no copies, You transfer all of the Application (including all component parts, media, printed or electronic materials, any upgrades and this Agreement) and the transferee agrees to the terms of this Agreement. If the Application is an upgrade, any transfer must include all prior versions of the Application. Single User License transfers are only allowed between Named Users of the same
4.2
organization (i.e., transfers are not allowed across different organizations). The Named User transferring the Application must deactivate any and all active installations of the Application before the transferee can Activate the Application. A Single User License may not be transferred more than two times in a calendar year. This section also applies to Single User License Rentals. Standard Network, Concurrent Network, Asset-based and Unlimited User Licenses. Standard Network, Concurrent Network, Asset-based and Unlimited User licenses are NOT transferable across organizations, and licensing rights CANNOT be split or transferred between multiple organizations in cases of corporate acquisitions or divestitures, except as allowed in items (a) and (b) of this section (Section 4.2), and with ReliaSoft Corporations written consent. In the case of Standard and Concurrent Network Licenses, Named User Activations can be deactivated and replaced by new users on a permanent basis. In other words, replacement of users is allowed to accommodate attrition and reassignment of user responsibilities in the ordinary course of doing business. However, replacing users on a daily, weekly, monthly or even quarterly basis and frequently reactivating users that were previously replaced is an indication that the number of Named Users (for Standard Network Licenses) or Named User Activations (for Concurrent Network Licenses) obtained by You is not adequate to serve the pool of users intending to Access the Application and is therefore prohibited. Each Standard Network License seat or Concurrent Network Named User Activation can be reassigned to a different Named User no more than two times in a calendar year. In cases of corporate acquisitions or divestitures: (a) Standard Network and Concurrent Network licenses remain with the organization that originally obtained the licenses, (e.g., if the organization that holds the license is acquired by another entity, the licensing rights will be transferred to the acquiring entity); (b) Unlimited Site-Wide licenses remain with the original licensing site and therefore pass on to the owner of the specific licensing site as long as the site continues to exist; (c) Unlimited Division-Wide licenses remain with the organization that originally obtained the licenses in cases where the Division is being partially divested, and are terminated when the Division is fully divested. As an example, in a partial divestiture, the portions of the Division divested will not have access to the Division-Wide license whereas the portions of the Division that remain will continue to have access to the Division-Wide license; and (d) Unlimited Corporate-Wide licenses remain with the organization that originally obtained the licenses as long as said organization divested a portion of its business or acquired other organizations, and are terminated in cases where said organization was acquired by another organization. As an example, in a partial divestiture, the portions of the corporation divested will not have access to the CorporateWide license whereas the portions of the corporation that remain will continue to have access to the Corporate-Wide license. As a further example, an organization with a Corporate-Wide license having acquired other entities (either in part or in whole) will continue to have the same licensing rights as before the acquisition and can therefore deploy the same licensing rights to the newly acquired entities. This section also applies to Standard Network and Concurrent Network License Rentals.
5 DESCRIPTION OF OTHER RIGHTS, LIMITATIONS AND MISCELLANEOUS ITEMS

5.1
5.2
5.3
5.4
5.5
Limitations on Reverse Engineering, Decompilation and Disassembly. You may not reverse engineer, decompile or disassemble the Application. You shall not provide, disclose or otherwise make available the Application, in any form, to any person other than your employees and under your direction and control for purposes specifically related to your permitted use of the Application. You will not: (a) alter, modify or prepare derivative works of the Application; (b) transmit the Application electronically by any means; or (c) cause or permit the translation, reverse engineering, disassembly, or decompilation of the Application to determine any design structure, source code, concepts and methodology behind the Application, whether to incorporate it within any product of your own creation, create a derivative work, create any product that is competitive with the Application or other ReliaSoft Corporation products, or for any other purpose. Copyright. All title and copyrights to the Application are owned by ReliaSoft Corporation (or its suppliers or licensors). The Application is protected by copyright laws and international treaty provisions. Therefore, You must treat the Application like any other copyrighted material except that You may make one copy of the media solely for backup or archival purposes. Proprietary Notices. All title, trademarks, copyrights and intellectual property rights in and pertaining to the Application (including but not limited to any copies thereof, software structure and organization, source code, images and new releases) are valuable property of ReliaSoft Corporation and are owned or licensed by ReliaSoft Corporation. You may not intentionally remove, alter or destroy any form of copyright and trademark notices, proprietary markings or confidential legends placed upon or contained within the Application, including but not limited to any such notices contained in physical and/or electronic media or documentation, in the Application interface boxes, or in any of the runtime resources, code or other embodiments originally contained in or dynamically or otherwise created by the Application. Use of ReliaSoft Corporations Marks. You may not use the name, logos, trade names or trademarks of ReliaSoft Corporation or any of its affiliates in any manner including, without limitation, in your advertising, promotional literature or any other material, whether in written, electronic, or other form distributed to any third party, except in the form provided by ReliaSoft Corporation and then solely for the purposes of identifying your use of the ReliaSoft Corporation Application. Verification. You will provide, on ReliaSoft Corporations reasonable written request, written verification that the Application is being used according to the terms of this Agreement. Upon thirty days prior written notice, if ReliaSoft Corporation has reasonable grounds to believe that this Agreement has been breached, ReliaSoft may audit your use of the Application provided such audit is: (a) limited to records relating solely to the Application necessary to verify compliance with the terms of this Agreement; (b) performed by a reputable independent third party auditor acceptable to You (acting reasonably); (c) the third party auditor shall comply with your standard security policies; (d) the third party
5.6
5.7 5.8 5.9
5.10
5.11
5.12
5.13
5.14
auditor shall execute your standard protective non-disclosure agreement; and (d) the cost of any requested audit will be solely borne by ReliaSoft Corporation if no breach is found as a result of the audit or will be solely borne by You if a breach is found. Such audit shall occur no more than once every twelve months and shall not unreasonably interfere with your normal business operations. Modification. ReliaSoft Corporation reserves the right to modify or enhance the Application without obligation to notify You of such changes or to furnish them to You, unless otherwise agreed upon with a separate agreement (such as an annual maintenance agreement). Copying. You may not, under any circumstances, copy the Application, in whole or in part, except as expressly provided under the Copyright section above. Separation of Components. The Application is licensed as a single product. Its component parts may not be separated for use on more than one computer. Rental or Other Exploitation. You may not publish, sub-license, re-license, assign, sell, distribute, license, transfer, rent, lease or lend the Application to any party, except transfer the Application as expressly provided under the Transfer section above. If you received any revenues from the unlawful distribution of the Application, such revenues will be forfeited to ReliaSoft Corporation. Fees. You will pay ReliaSoft Corporation all fees or other amounts due under this Agreement, plus any and all applicable taxes, within the payment term due date specified on the respective invoice. In the event that the respective invoice is not paid on time, or at all, ReliaSoft Corporation reserves the right to terminate this Agreement and revoke the corresponding licenses issued to You within the scope of this Agreement. Termination. Without prejudice to any other rights, ReliaSoft Corporation may terminate this Agreement if You fail to comply with the terms and conditions of this Agreement and such breach is not cured within thirty days of notice of such breach. In such event, You must destroy all copies of the Application and all of its component parts. Additionally, You may be held liable for any damage or loss of profit caused to ReliaSoft Corporation arising from unauthorized use or duplication of this Application. Supplemental Licenses. If, after the effective date of this Agreement, You subsequently purchase additional licenses of the Application, these supplemental licenses will be included under this Agreement. Press Releases. As part of this Agreement, You acknowledge that ReliaSoft Corporation may make reference to You as a customer of ReliaSoft Corporation in press releases, advertising and promotional materials, and You consent to any such reference. ReliaSoft Corporation will NOT disclose any further details beyond referring to You as a customer without prior written consent, not to be unreasonably withheld. Relationship. You and ReliaSoft Corporation are independent contractors and neither is an agent, joint venture partner, partner or employee of the other, and ReliaSoft Corporation will not be obligated by any agreements or representations made by You to any person, nor with respect to any other action by You, nor will ReliaSoft Corporation be obligated for any
5.15
5.16
5.17
5.18
damages to any person, whether caused by your actions, failure to act, negligence or willful conduct. Upgrades. If the Application is an upgrade from another product, whether from ReliaSoft Corporation or another supplier, You may use or transfer the Application only in conjunction with that upgraded product, unless You destroy the upgraded product. If the Application is an upgrade of a ReliaSoft Corporation product, You may use the upgraded product only in accordance with this Agreement. If the Application is an upgrade of a component of a package of software programs that You licensed as a single product, the Application may be used and transferred only as part of that single product package and may not be separated for use on more than one computer. U.S. Government Restricted Rights. The Application was developed at private expense. No portion of the Application was developed with government funds and the Application is a trade secret of ReliaSoft Corporation for all purposes of the Freedom of Information Act. The Application and documentation are provided with RESTRICTED RIGHTS. Use, duplication or disclosure by the Government is subject to restrictions as set forth in subparagraph (c)(1)(ii) of the Rights in Technical Data and Computer Software clause at DFARS 252.227-7013 (or its successor) or subparagraphs (c)(1) and (2) of the Commercial Computer Software Restricted Rights at 48 CFR 52.227-19 (or its successor), as applicable. Manufacturer is ReliaSoft Corporation, 1450 S. Eastside Loop, Tucson, Arizona 85710, USA. Limited Warranty. ReliaSoft Corporation warrants that: (a) the Application will perform substantially in accordance with the accompanying written materials, and on machines meeting the published minimum requirements, for a period of sixty (60) days from the date of receipt; and (b) any media on which the Application is furnished will be free from defects for a period of sixty (60) days from the date of receipt. Some states and jurisdictions do not allow limitations on duration of an implied warranty, so the above limitation may not apply to You; in such states and jurisdictions the liability of ReliaSoft Corporation shall be limited to the minimum extent permitted by law. To the extent allowed by applicable law, implied warranties on the Application and media, if any, are limited to sixty (60) days; and (c) at the time of installation, the Application will be free from any mechanism, feature or any other codes or instructions that: (i) cause the Application to remotely transmit information to ReliaSoft or any third party, except to communicate with one of ReliaSofts servers to facilitate specific functions of the Application (such as to communicate with the ReliaSoft License Server, access the online help files, etc.); or (ii) may be used to permit Access to, or use of, the Application or computer system on which the Application is loaded, or to which the Application is linked, by ReliaSoft or any third party. Customer Remedies. ReliaSoft Corporation's and its suppliers' or licensors entire liability and Your exclusive remedy shall be, at ReliaSoft Corporation's option, either (a) return of the fee paid for the Application, or (b) repair or replacement of the Application or media that does not meet ReliaSoft Corporation's Limited Warranty and which is returned to ReliaSoft Corporation with a copy of your receipt or invoice. Any replacement Application or media will be warranted for the remainder of the original warranty period or thirty (30) days,
5.19
5.20
5.21
5.22
whichever is longer. None of these remedies nor any product support services offered by ReliaSoft Corporation are available without a valid License Certificate issued by ReliaSoft Corporation. Warranty Exclusions. The Limited Warranty is void if the damage or defect has resulted from accident, abuse or misapplication. Any modification of the Application by any person other than ReliaSoft Corporation shall void this warranty. Any manipulation of the Applications data storage infrastructure or direct storage of data into the Applications data storage from outside the Application by any person other than ReliaSoft Corporation or ReliaSoft Corporations authorized representative shall void this warranty. The warranties in this section extend only to You and are contingent upon proper use of the Application. The warranties will not apply to any failure caused by (a) accident, (b) unusual physical, electrical or electro-magnetic stress, (c) negligence, (d) misuse, (e) failure of electrical power, air conditioning or humidity control, (f) use of the Application with any equipment or software not reflected in ReliaSoft Corporations specifications, (g) installation, alteration or repair of the Application by anyone other than ReliaSoft Corporation or ReliaSoft Corporations authorized representative, or (h) or installation on equipment on which the original identification marks have been removed or altered. No Other Warranties. No oral or written information or advice given by ReliaSoft Corporation, its suppliers, dealers, distributors or agents shall create a warranty or in any way increase the scope of the Limited Warranty, and You may not rely on any such information or advice as a warranty. Use of Results Provided By the Application Disclaimer. You understand that the results provided by the Application cannot replace judgment required for important decisions. Use of the results provided is done completely at your own risk. ReliaSoft Corporation does not warrant that the functions of this Application will meet your requirements or be error free. You assume all risk of the use, quality and performance of the Application, and You are advised to use your own discretion and judgment regarding the use of the Application. RELIASOFT CORPORATION, ON BEHALF OF ITSELF AND ITS LICENSORS, DISCLAIMS ALL OTHER WARRANTIES, EITHER EXPRESS OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE, WITH REGARD TO THE APPLICATION. IN NO EVENT SHALL RELIASOFT CORPORATION OR ITS SUPPLIERS OR LICENSORS BE LIABLE FOR ANY SPECIAL, INCIDENTAL, INDIRECT, OR CONSEQUENTIAL DAMAGES WHATSOEVER (INCLUDING, WITHOUT LIMITATION, DAMAGES FOR LOSS OF BUSINESS PROFITS, BUSINESS INTERRUPTION, LOSS OF BUSINESS INFORMATION OR ANY OTHER PECUNIARY LOSS) ARISING OUT OF THE USE OF OR INABILITY TO USE THE APPLICATION, EVEN IF RELIASOFT CORPORATION HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. USE OF THIS APPLICATION IS DONE COMPLETELY AT YOUR OWN RISK, INCLUDING THE USE OF RESULTS PROVIDED BY THIS APPLICATION. RELIASOFT CORPORATIONS LIABILITY HEREUNDER SHALL BE LIMITED TO PHYSICAL DAMAGE DIRECTLY CAUSED BY THE SOLE NEGLIGENCE OF RELIASOFT CORPORATION AND SHALL NOT EXCEED THE PRICE
5.23
5.24
5.25 5.26
5.27
5.28
PAID FOR THE SOFTWARE PRODUCT. NOTWITHSTANDING ANYTHING HEREIN TO THE CONTRARY, THE FOREGOING DISCLAIMER AND LIMITATION OF LIABLITY SHALL NOT APPLY TO RELIASOFT CORPORATIONS OBLIGATIONS UNDER ANY CLAIM OF INFRINGEMENT AS DESCRIBED IN SECTION 5.28 OF THIS AGREEMENT. Venue. Venue for any proceedings arising out of or relating to this Agreement shall be in Tucson, Pima County, Arizona. The parties disclaim the application of the United Nations Convention on Contracts for the International Sale of Goods. This Agreement is governed by the laws of the State of Arizona, USA, without reference to conflict to law principles. Each party to this Agreement submits to the exclusive jurisdiction of the state and federal courts in the State of Arizona for the purpose of resolving any disputes arising under or relating to this Agreement. Each party waives any jurisdictional, venue or inconvenient forum objections to such courts. Legal Expenses. If legal action is taken by either party to enforce this Agreement, all costs and expenses (including reasonable attorney fees) incurred by the prevailing party in exercising any of its rights or remedies or in enforcing any of the terms, conditions, or provisions of this Agreement will be paid by the other party. Force Majeure. ReliaSoft Corporation will not be responsible for delays or failures in its performance due, in whole or in part, to any cause beyond its reasonable control. Export Control. Regardless of any disclosure made by You to ReliaSoft Corporation of an ultimate destination of the Application, You will not export, either directly or indirectly any Application without first obtaining all licenses required, from the U.S. Department of Commerce or any other agency or department of the United States Government, and complying with the applicable laws. Neither the Application nor any direct product thereof may be exported, directly or indirectly, in violation of applicable export laws, or may be used for any purpose prohibited by these laws including, without limitation, nuclear, chemical or biological weapons proliferation. ReliaSoft Corporation will provide You with all reasonable information requested by You in connection to exporting the Application, including providing You with the U.S. Export Controls Classification Number (ECCN) for the Application. Waiver. The waiver by either party of any breach of this Agreement shall be in writing and shall not constitute a waiver of any other or subsequent breach. No waiver of any of the provisions of this Agreement will be deemed, or will constitute, a waiver of any other provision, whether or not similar, nor will any waiver constitute a continuing waiver. The failure by a party to enforce any provision of this Agreement will not be deemed a waiver of future enforcement of that or any other provision. Indemnification. You will indemnify and hold ReliaSoft Corporation harmless against any and all claims, damages, losses, costs or other expenses (including reasonable attorney fees) that arise directly or indirectly from your breach of this Agreement. ReliaSoft Corporation shall defend, indemnify and hold harmless, at its own expense, You and your assigns, successors, directors, officers and employees (each an Indemnified Party) against any and all claims incurred by or made against an Indemnified Party by a third party in connection with a claim, suit or action which is based on an allegation that the Application when used by You as authorized under this Agreement, misappropriates or infringes any
5.29
5.30
third party patent, copyright, trade secret or other intellectual property right (each, a Claim) provided that ReliaSoft Corporation shall have received from the Indemnified Party: (i) notice of such Claim as soon as possible after You receive notice of the Claim; given that a failure to provide notice shall only relieve ReliaSoft Corporation of its indemnity obligation to the extent ReliaSoft Corporation was prejudiced by such failure; (ii) the exclusive right to control and direct the investigation, defense or settlement of such claim; and (iii) all reasonable necessary cooperation by You. If your use of any of the Application is, or in ReliaSoft Corporations opinion is likely to be, enjoined due to a Claim, ReliaSoft Corporation may, at its sole discretion: (a) modify the Application so that it becomes noninfringing, provided such modifications result in software with substantially similar functionality and performance; (b) procure for You the right to continue using the Application under substantially the same terms and conditions as provided for hereunder; or (c) if (a) and (b) are commercially impracticable, terminate the Agreement and refund to You the license fee paid by You for the Application which is the subject of the Claim as reduced to reflect a three-year straight-line depreciation from the applicable license purchase date. The foregoing indemnification obligation of ReliaSoft Corporation shall not apply: (1) if the Application is modified by any party other than ReliaSoft Corporation and such modification was not authorized in writing by ReliaSoft Corporation, but solely to the extent the alleged infringement is caused by such modification; or (2) to any release of the Application other than the most current release, provided that: (I) the most current release was either made available at no cost to You and (II) You had a commercially reasonable period of time (not to exceed 60 days) after availability of the current release to implement the current release so as to avoid the infringement claim. This section (Section 5.28) sets forth ReliaSoft Corporations sole liability and your sole and exclusive remedy with respect to any claim of infringement. Equitable Relief. You acknowledge and agree that, due to the unique nature of the Application, there can be no adequate remedy at law for any breach of your obligations under this Agreement, that any such breach may allow You or third parties to unfairly compete with ReliaSoft Corporation resulting in irreparable harm and therefore that, upon any such breach or threat thereof, ReliaSoft Corporation shall be entitled to injunctive and other appropriate equitable relief in addition to whatever remedies it may have at law. Entire Agreement; Amendments. This Agreement is the complete and exclusive statement of the agreement between the parties and supersedes all prior agreements and communications with respect to the subject matter, and there are no oral representations, understandings or agreements that are not fully expressed herein. Any terms appearing on any order or other form used by You which modify or conflict with the terms and conditions set forth herein are expressly rejected. Except for the purpose of negating implied warranties, no course of prior dealings between the parties and no usage of the trade shall be relevant to supplement or explain any term used in this Agreement. No ReliaSoft Corporation employee other than an officer of ReliaSoft Corporation (Vice President and above) shall have any actual or apparent authority to modify the terms of this Agreement in any way. All amendments shall be in writing and signed by the authorized representative of ReliaSoft Corporation.
5.31
5.32
Severability. If any one or more of the provisions of this Agreement shall for any reason be held to be invalid, illegal or unenforceable in any respect, any such provision shall be severable from this Agreement, in which event this Agreement shall be construed as if such provision had never been contained herein. Electronic Signatures. ReliaSoft Corporation and You agree that this Agreement may be executed electronically and that electronic copies of this Agreement shall be binding upon the parties to the same extent as manually-executed copies.

Qs Weibullalta

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Qs Weibullalta

Uploaded by

Copyright:

Available Formats

Use with Quick Start Repository: Revision 1.

Quick Start Guide

Standardfolios Completeand censoreddata Confidence intervals QCP,plotsand reports

Similarly, you use the QCP to answer the remaining questions:

2-sided confidence bounds

calculated results are shown next.

as shown in the picture on page 22.

Degradation folios Analyzing extrapolated failuretimes QCPandplots

Next, you click the Calculate icon to analyze the data.

Y-axis: 450 to 1,000 lumens X-axis: 300 to 3,000 hours

Warrantydata analysis Nevadachart format Forecasting productreturns

The following picture shows the 2P-Weibull parameters of bulb A.

Estimatingtarget reliability Calculatingthe returnonan investmentin reliability

Here is a more detailed breakdown of the inputs:

The plot appears as shown next.

StressStrength tool Reliability Demonstration Test(RDT)tool

Value: Required test time Sample Size: 10 Maximum Number of Failures: 0

Competingfailure modesanalysis Batchautorun Reliabilityblock diagrams

The result will appear, as shown next.

Recurrentevent dataanalysis ParametricRDA folios

8 825 847 939 995 1209 1500

10 659 913 916 1309 1319 1500

15 449 582 898 1101 1154 1334 1500

17 614 784 921 1100 1298 1475 1500

22 616 791 792 923 1413 1500

23 657 717 762 909 1041 1277 1444 1500

24 310 726 868 1009 1017 1292 1310 1500

25 415 559 847 982 986 1378 1500

26 578 816 874 1082 1148 1434 1483 1500

27 832 917 1024 1048 1463 1500

The Results window is shown next.

Tip:Inthedatasheetnames,the[F]tagindicatesananalysisofthefailures,while[E]isananalysisofthegeneral events(notethatacombinedanalysiswoulddisplay[C]).The[FD]tagindicatesthatthedatasetisforthefailure distribution,while[RD]isthedatasetfortherepairdistribution.

ALTAwithone stresstype ALTAQuick CalculationPad (QCP)

ALTAwithtwo stresstypes ALTAQCPand plots

You use the QCP to solve the remaining metrics:

The following picture shows the completed setup.

You continue using the QCP to solve the remaining metrics:

LuminosityMeasurementsat138Volts Inspection Time(Hr)

To analyze the data set, you click the Calculate icon.

RELIASOFT CORPORATION END-USER LICENSE AGREEMENT

1.3 1.4 1.5

INSTALLATION AND USE

5 DESCRIPTION OF OTHER RIGHTS, LIMITATIONS AND MISCELLANEOUS ITEMS

5.7 5.8 5.9

You might also like